POLYIMIDES FOR ADVANCED APPLICATIONS by David C. Rich S.B., Chemical Engineering Massachusetts Institute of Technology, 1992 Submitted to the Department of Materials Science and Engineering in Partial Fulfillment of the Requirements for the Degree of DOCTOR OF PHILOSOPHY in Polymers at the Massachusetts Institute of Technology May 3, 1996 @Massachusetts Institute of Technology. All rights reserved. Signature of Author..................................... ....... David Rich Department of Materials Science and Engineering Certified bv............................ Professor Peggy Cebe -m,--4i Qipervisor Accepted by...... ................................................ rilc r- ............. ui iviiael Rubner Chairman, Departmental Committee on Graduate Studies I1NS'."• SE'Tq'S ,,;AG.SACH-Ui •.L OF TECHNOLOGY JUN 2 41996 Science UBRPARiES POLYIMIDES FOR ADVANCED APPLICATIONS by DAVID C. RICH Submitted to the Department of Materials Science and Engineering on May 3, 1996 in partial fulfillment of the requirements for the Degree of Doctor of Philosophy in Polymers Thesis Supervisor: Professor Peggy Cebe ABSTRACT The processing, structure, and properties of aromatic polyimides are investigated for their use as liquid crystal alignment layers and in other advanced applications. Probimide 32 polyamide-imide and Probimide 412 inherently photosensitive polyimide are the primary materials which are evaluated. The physical and chemical changes that occur in the polyimide films as the materials undergo cloth brushing, UV irradiation, and thermal curing treatments are of chief interest. Polyimide films are studied by conventional characterization techniques, by property prediction methods, and by a new technique called "alignment layer relaxation," which can be used to detect thermal transitions in polymer surface layers. The alignment relaxation technique is based upon the observation that as alignment layers are cured above their glass transition temperatures, their liquid crystal aligning properties are irreversibly destroyed. The aligning performance of polymer films is evaluated by the phase-sensitive laser optical transmission of parallel-aligned nematic liquid crystal cells placed between crossed polarizers. A non-contact method of alignment of surface aligning liquid crystals is also introduced, in which polyamide-imide alignment films are irradiated with linearly polarized short wavelength ultraviolet light (,=254 nm). The technique produces alignment that appears to be comparable with that achieved by conventional contact brushing. The photo-aligned polyamide-imide is believed to develop its aligning properties via a chain scission process that occurs in the alignment films as a consequence of the UV exposure, though the exact mechanism of the alignment is not completely understood at this time. The removal of solvent from polyimide films upon 3 curing is investigated by thermogravimetric analysis and Fourier transform infrared spectroscopy. Solvent is found to be rapidly eliminated from polyamide-imide thin films (-450A) at low temperatures (-110 0C). High temperature cures are found to chemically crosslink the polyimides, especially if the treatment proceeds in an air environment. Also, comparison of predicted and measured refractive indices of polyimides materials indicate that, upon thermal imidization, non-fluorinated polyimides undergo chain packing and/or charge transfer processes that appear to substantially mitigated in polyimides with bulky fluorinated spacer groups. Table of Contents Abstract ......................................................................................................................... 2 Table of Contents................................................................................................... List of Figures........................................................................................................ List of Tables................................................................................................................. Acknowledgments ......................................................................................................... 4 9 19 21 Chapter 1. Introduction and Theory.................................................................. 22 1.1. Synthesis and Processing............................ 1.1.1. General scheme....................................................................................... 1.1.2. Photosensitive polyimide precursors....................................................... 1.1.3. Preimidized polyimides............................................. 1.1.4. Preimidized, photosensitive polyimides............................. ........ 22 22 24 24 26 1.2. Structure ............................................................................................................... 1.2.1. Macroscopic orientation of polymer films............................................... 1.2.2. Crystalline packing ................................................. 1.2.3. Chain-chain interactions................................................................ 1.2.4. Liquid crystalline nematic order and alignment.............................. 26 26 26 28 29 1.3. Properties............................................................................................................. 1.3.1. Quantitative relationship between polymer chemistry and refractive index................................................................. 1.3.2. Property changes associated with charge transfer complexation............................................. 1.3.3. Anisotropy of optical properties.............................................................. 1.3.4. Measurement of index of refraction................... .......... 1.3.5. Optical transmission model for liquid crystal alignment.......................... 30 30 37 39 45 46 1.4. Performance: Liquid Crystal Alignment Layers............................... ....... 53 1.4.1. Liquid crystal surface alignment on brushed polymer films.......................................................................................... 53 1.4.2. Alternative treatments to polymer layers for liquid crystal alinnment...... .......................... cr st l ale m n . .. .... ............... ........................... 1.4.3. An LCD device: the TN-LCD............................ ... ............... 58 1.5. References............................................................................................................ 60 Chapter 2. Refractive Indices of Aromatic Polyimides................................. ....... 72 2.1. Introduction ......................................................................................................... 72 2.2. Experimental Techniques................................................. 2.2. 1. Polyimide film preparation....................................... 2.2.2. Zone drawing .......................................................................................... 2.2.3. Refractive index measurement ......................... .................................. 2.2.4. Refractive index prediction...................................................................... 74 74 74 75 76 2.3. Relationship Between Polyimide Chemistry and Refractive Index....................... 2.3.1. Amorphous, non-fluorinated polyimides............................ ....... 2.3.2. Semicrystalline polyimides .......................................... 2.3.3. Polyimides containing hexafluoroisopropylidene (6F) ............................ 2.3.4. Polyimides containing fluorinated alkoxy side chains ............................. 2.3.5. Polyimides containing pentafluorosulfanyl (SF5)................................... 76 76 77 77 78 79 2.4. Relationship Between Polyimide Processing and Refractive Index................... 2.4.1. In-plane birefringence in polyimide films.................................. ...... 2.4.2. Birefringence in zone drawn films.................................. ......... 2.4.3. Effect of cure temperature on refractive index......................................... 2.4.4. Effect of post-cure annealing on NEW-TPI polyimide ............................ 89 89 90 91 96 2.5. Conclusions.......................................................................................................... 96 2.6. References............................................................................................................ 99 Chapter 3. Curing Study of Probimide® 412 Preimidized Photosensitive Polyimide........................................................................... 104 3.1. Introduction ......................................................................... ........................... 104 3.2. Experimental........................................................................................................ 3.2.1. Polymer film samples................................. 3.2.2. Measurement techniques...... ...................... 105 105 106 3.3. Results and Discussion............................... 3.3.1. Coloration and brittleness ............................. 3.3.2. Thermogravimetric analysis.............................. 3.3.3. Dynamic mechanical analysis.................................................................. 3.3.4. Index of refraction............................. 3.3.5. Fourier transform infrared spectroscopy ................................................. 108 108 108 116 124 124 3.4. Conclusions .......................................................................................................... 128 3.5. References...................... .............................................................................. 139 Chapter 4. Effect of Cure Conditions on Probimide® 32 Polyamide-imide........................................................................................ 141 4. 1. Introduction.......................................................................................................... 141 4.2. Experimental......................................................................................................... 4.2.1. Polymer film processing..................................................................... 4.2.2. Measurement techniques .............................. 142 142 143 4.3. Results and Discussion.............................. 4.3.1. Coloration and brittleness ............................. 4.3.2. Thermogravimetric analysis .................................................................... 4.3.3. Dynamic mechanical analysis.................................................................. 4.3.4. Index of refraction.................................................. ............................ 4.3.5. Fourier transform infrared spectroscopy............................ 143 143 144 152 153 160 4.3.6. Differential scanning calorimetry............................................................. 167 4.4. Conclusions .......................................................................................................... 185 4.5. References............................................................................................................ 185 Chapter 5. Alignment of Liquid Crystals on Brushed Polymer Layers.................................................................................. 187 5.1. Introduction.......................................................................................................... 187 5.2. Experimental Section....................................................... 5.2.1. Construction of LC alignment cells ......................................................... ................. 5.2.2. Evaluation of alignment cells................................... 188 189 195 5.3. Results and Discussion............................ 5.3.1. Brushing-induced alignment in Probimide 32 polyamide-imide............................................................... 5.3.2. Brushing-induced alignment in other polymers................................. 5.3.3. FTIR spectra of Probimide 32 polyamide-imide alignment films................................ 5.3.4. Atomic force microscopy of brushed polyamide-imide alignment films............................ 5.3.5. In situ thermal stability of LC alignment on brushed Probimide 32 polyamide-imide layers................. ................. 5.3.6. "Liquid crystal alignment relaxation" - a new technique for measuring thermal transitions in surface layers............................ 205 5.4. Experimental Error.................................... 227 5.5. Conclusions .......................................................................................................... 228 5.6. References ........................................................................................................... 229 205 207 209 210 216 217 Chapter 6. Liquid Crystal Surface Alignment on Photoirradiated Polymer Layers.............................................. 232 6.1. Introduction.......................................................................................................... 232 6.2. Experimental Section....................................................... 6.2.1. Photo-alignment technique............................ .................. 6.2.2. Comments regarding the methodology by which the PAI was selected.................................................... 232 232 6.3. Results and Discussion.................................................... 6.3.1. Evidence of liquid crystal alignment upon LP-UV exposure................. 6.3.2. Effect of unpolarized UV exposure............................. .............. 6.3.3. Effects of linearly polarized UV exposure................................................ 234 234 237 240 6.4. Conclusions.......................................................................................................... 250 6.5. Subjects for further work..................................................................................... 250 6.6. References ...................................................................................................... 252 Appendix A. Description of Nematic Alignment in Terms of the Poincare Sphere .................................. 253 A. 1. The Poincare sphere................................... 253 A.2. Changes in polarization state of the Poincare sphere............................... 254 A.3. Examples of nematic LC orientation............................... 257 A.4. Effect of misalignment of the LCs............. 259 ................... A.5. References ...................................................................................................... 234 259 List of Figures Figure 1.1. General synthesis of polyimides............................................................ 23 Figure 1.2. Processing scheme of photosensitive polyimide precursors............... 25 Figure 1.3. Types of orientation in polymer films. a.) isotropic orientation, b.) in-plane orientation, and c.) uniaxial orientation.............................. Figure 1.4. Charge transfer interactions in polyimides........................... Figure 1.5. A nematic liquid crystalline phase with director D' ............................... Figure 1.6. Representations of nematic liquid crystalline material. a.) no alignment and b.) bulk alignment along director D'.................. Figure 1.7. Sample calculation of the index of refraction of polystyrene using the van Krevelen / Vogel technique......................... Figure 1.8. Bicerano prediction for the refractive index of polystyrene................ 38 Figure 1.9. Examples of optical anisotropy. a.)birefringence on the molecular level, b.) birefringence on the microscopic level in a polymer spherulite, c.) birefringence on the microscopic level in a nematic grain, d.) bulk birefringence in an in-plane oriented film, and e.) bulk birefringence in a film of aligned liquid crystals.................................................. 41 A thick semicrystalline polymer block that is microscopically birefringent but has nx=ny=nz................................................................ 44 Figure 1.10. ...... 28 29 Figure 1.11. Prism coupling technique for measuring index of refraction............... 45 Figure 1.12. Graphical representation of a parallel-aligned liquid crystal plate between a polarizer (p) and an analyzer (A)............................... 46 Figure 1.13. Plot of I(0)/I(450) vs ................................................... 48 Figure 1.14. Elliptically polarized light with handedness (left or right), azimuth (a), and ellipticity (13= a/b)................................. 49 ....... Figure 1.15. Parallel-aligned liquid crystals with inhomogeneities.............. 50 Figure 1.16. Rays of light passing through different points 1 or 2 of the inhomogenous liquid crystal sample and emerging at 1' or 2' with different polarization states................................ 51 ...... Figure 1.17. Alignment of liquid crystals on brushed polymer surfaces................. 53 Figure 1.18. A polymer surface being rubbed by a brushing wheel........................... 54 Figure 1.19. The photochemical reaction in polyvinylmethoxycinnamate and the oriented structure, proposed by Schadt, that is formed by exposure to linearly polarized UV........................................ 56 Anisotropic chain degradation proposed to occur in a polyimide film during exposure to linearly polarized UV ....................................... 57 States of a twisted nematic liquid crystal display (TN-LCD) device, a.) off state and b.) on state................................. 59 Figure 1.20. Figure 1.21. ....... Figure 2.1. Zone drawing method.................................................................. 75 Figure 2.2. Experimental and predicted refractive indices of non-fluorinated polyimides in Table 2.1.................................. ..... .............. 79 Figure 2.3. Wide angle X-ray scattering (WAXS) of: a.)PMDA-ODA and b.) ODPA-PDA...................................................... Figure 2.4. Measured and predicted refractive indices of semicrystalline polyimides in Table 2.2......................................................................... 82 Figure 2.5. Measured and predicted refractive indices of 6F polyimides in Table 2.3......................................................... Figure 2.6. Predicted refractive indices (Bicerano's method) of polyimides with fluorinated alkoxy side chains..................................................... Figure 2.7. Refractive index vs. draw ratio for 6FDA-4BDAF in the direction of the draw and normal to the draw............................. Figure 2.8. Refractive indices of several polyimides as a function of cure temperature............................................ Figure 2.9. Predicted (Bicerano's method) vs. measured refractive indices of several polyimides. a.) Predicted values of the polyamic acid precursors vs. measured values of the films after only 100 0C cure and b.) Predicted values of the polyimides vs. measured ..................... 94 values after 3000 C cure................................ Figure 2.10. a.) Average refractive index of NEW-TPI vs. post-cure annealing temperature and b.) Birefringence vs. post-cure annealing temperature ............................................................................ 97 Figure 2.11. a.) Average refractive index vs. post-cure annealing time at 2900 C and b.) Birefringence vs. post-cure annealing time at 2900C........................................................... 98 Figure 3.1. Preparation of bulk films. a.) Coating and blading process for optimization of film smoothness and thickness uniformity. b.) Pre-cure air-drying chamber.............................. 107 Thermogravimetric weight loss and derivative of weight loss vs. temperature for Probimide® 412 thick films with highest cure temperature: (a) room temperature, (b) 100 0C, (c) 140 0 C, (d) 2000 C, and (e) 300 0C................................ 110 Figure 3.2. Figure 3.3. Figure 3.4. Figure 3.5. Figure 3.6. Thermogravimetric weight loss vs. temperature for Probimide® 412 thick films baked at 100 0C, UV irradiated or unirradiated ..................................................................................... ...... 115 1 Hz dynamic Young's modulus and loss factor, tanS, vs. temperature for Probimide® 412 thick films with highest cure temperature: (a) 100 0C, (b) 2000C, (c) 300 0C (in nitrogen), (d) 400 0C (in nitrogen) ........... ........................... 118 1 Hz dynamic Young's modulus and loss factor, tans, vs. temperature for Probimide® 412 thick films cured at 100 0 C, UV irradiated (dashed line) or unirradiated (solid line) .......................... 122 1 Hz dynamic Young's modulus and loss factor, tan6, vs. temperature for Probimide® 412 thick films cured at 300 0C, in air (dashed line) or in nitrogen (solid line)..................................... 123 Figure 3.7. Indices of refraction of spin coated Probimide® 412 films as a function of highest cure temperature.......................... ............. 125 Figure 3.8. Indices of refraction of spin coated Probimide 412 films as a function of UV exposure time................................. ............... Figure 3.9. 126 Relative IR absorbance vs. wavenumber for spin coated Probimide® 412 films softbaked 15 minutes at 100 0C and exposed to UV: (a) 0 minutes (no irradiation), (b) 2 minutes, (c) 10 minutes, (d) 50 minutes........................................... .......... 130 Figure 3.10. Relative IR absorbance vs. wavenumber (2000-4000 cm -1 range) for spin coated Probimide® 412 films with highest cure temperature treatment: (a) 300 0C (3 hours in nitrogen), (b) 3000 C (1 hour in air), (c) 3000 C (3 hours in air), (d) 4000C (1 hour in nitrogen).............. ........................... 134 Figure 3.11. Figure 4.1. Figure 4.2. Relative IR absorbance vs. wavenumber (400-2000 cm -1 range) for spin coated Probimide® 412 films with highest cure temperature treatment: (a) 3000C (3 hours in nitrogen), (b) 3000 C (1 hour in air), (c) 3000 C (3 hours in air), ....... (d) 4000 C (1 hour in nitrogen).............. .......... 135 Weight loss and derivative of weight loss vs. temperature for N-methyl pyrrolidone (NMP)............................... 145 Weight loss and derivative of weight loss vs. temperature for Probimide® 32 thick films: a.) dried under an air current at room temperature, b.) then baked 6 hours at 100 0 C, c.) then 3 hours at 2000C, d.) then 3 hours at 300 0C...................................... 146 Figure 4.3. Weight loss and derivative of weight loss vs. temperature for Probimide® 32 thick films baked at 100 0 C for 1, 6, or 24 hours.......... 151 Figure 4.4. Young's modulus and dissipation factor, tan5, for Probimide® 32 thick films: a.) cured 6 hours at 100 0C, b.) then 3 hours at 2000 C, c.) then 3 hours at 3000 C in either air or, d.) nitrogen............ 154 Figure 4.5. a.) Indices of refraction of spin coated Probimide® 32 films vs. highest cure temperature in air. Cure cycle was 15 minutes at 100 0C, 1 hour at 200 0C, 1 hour at 3000 C. b.) Refractive indices for different cure conditions at 300 0C................................ 158 Relative absorbance vs. wavenumber (400-2000 cm-1) for spin coated Probimide® 32 films: a.) baked 10 min. at 100 0C, b.) 10 min. at 120 0C, c.) 10 min. at 140 0C, d.) 10 min. at 160 0 C, e.) 10 min. at 180 0C, f.) 10 min. at 2000C ............................................ 168 Relative absorbance vs. wavenumber (2000-4000 cm- 1) for spin coated Probimide® 32 films: a.) baked 10 minutes at 100 0C, b.) then 10 minutes at 120 0C, c.) then 10 minutes at 140 0C, d.) then 10 minutes at 160 0C, e.) then 10 minutes at 180 0C, f.) then 10 minutes at 200 0C................................ 174 Figure 4.6. Figure 4.7. Figure 4.8. Figure 4.9. Figure 4.10. Relative absorbance vs. wavenumber (400-2000 cm- 1) for spin coated Probimide® 32 films: a.) baked 15 minutes at 100 0C, b.) then 1 hour at 2000 C, c.) then 1 hour at 300 0 C in either air or, d.) nitrogen ................................ 175 Relative absorbance vs. wavenumber (2000-4000 cm- 1) for spin coated Probimide® 32 films: a.) baked 15 minutes at 100 0 C, b.) then 1 hour at 2000 C, c.) then 1 hour at 3000 C in either air or, d.) nitrogen ................................ 179 Ratio of relative absorbance of peak at 1680 cm- 1 to that of the 1370 cm -1 vs. highest cure temperature.......................... 180 Figure 4.11. Endothermic heat flow vs. temperature for Probimide® 32 thick films: a.) baked 6 hours at 100 0C, b.) then 3 hours at 2000 C, c.) then 3 hours at 3000 C in either air or, d.) nitrogen........................ 181 Figure 5.1. Alignment film brushing technique............................ 191 Figure 5.2. Cell constuction procedure. (a.) alignment film formation, (b.) empty cell construction, (c.) cell filling, (d.) cell annealing, and (e.) cell sealing ................................................... 193 Qualitative observations of liquid crystal alignment cells between crossed polarizers..................................................................... 196 Figure 5.3. Figure 5.4. Optical transmission apparatus with phase-sensitive detection........... 198 Figure 5.5. Instrument connections for phase-sensitve detection.......................... 199 Figure 5.6. Measure relative intensity vs. expected relative intensity for phase-sensitive detection of neutral density filtered light. a.) linear scale, and b.) log-log scale................................ 201 Figure 5.7. Measure relative intensity vs. expected relative intensity for phase-sensitive detection of light diminished by two consecutive polarizers with their polarizing axes lying at an angle X. a.) linear scale, and b.) log-log scale.............................. 202 Figure 5.8. Adjustment made for a cell with small degree of twist distortion q. The first liquid crystal surface director N is oriented parallel to polarizer P. The second surface director is N'. The analyzer is adjusted by q from A to A' such that A' is perpendicular to N'...................................... 204 Figure 5.9. a.) Intensity / intensity at 450 vs. rotation angle for a nematic liquid crystal cell aligned with brushed Probimide 32 polyamide -imide layers. The x-axis is the azimuth of the alignment director with respect to the polarizer. b.) Relative intensity vs. rotation angle for nematic liquid crystal cell with untreated Probimide 32 polyamide-imide layers. The x-axis represents the azimuth of an arbitrary reference axis with respect to the polarizer................................................................................................. 206 Figure 5.10. Values of 4, the effective angle of misalignment, for liquid crystal cells produced with alignment films of brushed PMDA-APB polyimide (PI), PMDA-APB poly(amic acid) (PAA), polyamide-imide (PAI), poly(phenylene ether sulfone) (PPES), poly(vinyl alcohol) (PVA), poly(methyl methacrylate) (PMMA), and polycarbonate (PC). All cells but those produced with PMMA and PC appear to have very strong macroscopic alignment ............................. . 209 Relative absorbance vs. wavenumber for Probimide 32 polyamide-imide coated from 3wt% solids content and softbaked at 110C, with final film thickness approximately 450A ...................................................................................................... 211 Figure 5.11. Figure 5.12. Figure 5.13. Relative absorbance vs. wavenumber for a brushed Probimide 32 polyamide-imide thin film. a.) Incident infrared electric field polarized parallel to the brushing direction. b.) Incident infrared electric field polarized perpendicular to the brushing direction .............................. 212 Atomic force microscopy (AFM) images of polyamideimide films. a.) brushed, and b.) unbrushed........................................ 214 Figure 5.14. 1(00)/1(450) vs. temperature for a polyamide-imide aligned liquid crystal cell undergoing an annealing cycle of systematically increasing temperature treatments................................... 216 Figure 5.15. 1(00)/1(450) vs. post-brush cure temperature for LC alignment cells produced with Probimide 32 polyamide-imide alignment films.......... 218 Figure 5.16. Depiction of the structure of the alignment layer films during alignment relaxation process of the polyamide-imide. The amorphous chains begin randomly coiled in the plane of the film (a). They are oriented by the brushing process (b). The orientation is reversed by annealing above Tg (c)............................... 219 Figure 5.17. 1(00)/1(450) vs. post-brush cure temperature for liquid crystal Figure 5.18. Figure 5.19. alignment cells produced with brushed polyamide-imide that were either softbaked (filled circles) or cured at 300 0C in air (open circles) prior to brushing............................. 220 1(00)/1(450) vs. post-brush cure temperature for liquid crystal alignment cells produced with brushed poly(phenylene ether sulfone) alignment films................................ 221 1(00)/1(450) vs. post-brush cure temperature for liquid crystal alignment cells produced with brushed PMDA-APB alignment films. a.) PMDA-APB poly(amic acid) softbaked only, b.) PMDA-APB hard baked 275 0C 1 hour, and c.) PMDA-APB cured at 221 0 C for 30 minutes.............................. 223 Figure 5.20. Figure 6.1. Figure 6.2. Figure 6.3. vs. post-brush cure temperature for liquid crystal alignment cells produced with poly(vinyl alcohol) alignment layers................................... 226 Exposure technique for irradiating alignment films with linearly polarized ultraviolet light (LP-UV)............................ 233 I(0)/I(450) vs. rotation angle for liquid crystal alignment cell produced with polyamide-imide alignment layers that have been irradiated with LP-UV for 25 minutes.............................. 235 1(00)/1(450) 1(00)/1(450) vs. LP-UV exposure time for liquid crystal cells produced with LP-UV irradiated polyamide-imide alignment films.......................................................... 236 Figure 6.4. Figure 6.5. Relative absorbance vs. wavenumber for a Probimide 32 polyamide-imide film. a.) unexposed to UV, and b.) UV exposed................................................................................................. 238 1(00)/1(450) vs. post-brush curing temperature for liquid crystal cells produced with UV exposed (squares) and unexposed (circles) polyamide-imide alignment layers................ 240 Figure 6.6. Relative absorbance vs. wavenumber for polyamide-imide layers exposed to linearly polarized ultraviolet (.=245 nm). a.) 2 hour exposure, UV axis parallel to incident IR axis, b.) 2 hour exposure, UV axis perpendicular to incident IR axis, c.) 4 hour exposure, UV axis parallel to incident IR axis, and d.) 4 hour exposure, UV axis perpendicular to incident IR axis............ 242 Figure 6.7 I(00)/1(450) vs. post-aligning cure temperature for liquid crystals cells produced with polyamide-imide alignment films that have been exposed to LP-UV (triangles) or brushed (circles)......... 246 Figure 6.8. Atomic force microscopy (AFM) image of LP-UV irradiated polyamide-imide films........................................................................... 247 Figure 6.9. 1(0')/1(450) vs. post-aligning unpolarized UV exposure time for liquid crystal cells produced from polyamide-imide alignment films that were exposed to LP-UV (open circles) or brushed (filled circles) .............................. 249 Figure A.I. Lines of latitude (1)on the Poincare sphere representing ellipticity ([3)........... ........................................................................ 253 Figure A.2. Lines of longitude (L) on the Poincare sphere representing azimuth (a) ..................................... 254 Movement along the Poincare sphere from reference point P along the small circle of the sphere centered on OS, where O is the center of the sphere and S represents the azimuthal angle parallel to the fast vibration axis of the crystal.............................. 255 Figure A.3. Figure A.4. The relative change in polarization state induced by the analyzer is a longitudinal shift through the arc length between the analyzer (A) and the point of reference (P). X1 moves to X2......................... 256 Figure A.5. The arc length from reference state P1 to final state P3 is given by 2p. The intensity transmitted by the system is cos 2p where full intensity is unity ........................................ 257 Figure A.6. Parallel aligned LC plate with nematic vector parallel to P.................. 258 Figure A.7. LC cell with nematic vector 450 with respect to P and A. Figure A.8. The light is elliptically polarized by the sample as it moves along the prime meridian to point E.................................... 258 Effect of random inhomogenous region on the Poincare sphere analysis. The shaded region represents a distribution of polarization states deviating from A ....................................................... 259 List of Tables Table 1.1. Meaning of the ratio of transmitted intensity at 00 (MIN) to transmitted intensity at 450 (MAX)................................ ....... 52 Table 2.1. Dianhydride (R) and diamine (Z) elements of polyimides in Figure 2.2....................................................................................... 81 Table 2.2. Dianhydride (R) and diamine (Z) elements of polyimides in Figure 2.4....................................................... 83 Dianhydride (R) and diamine (Z) elements of the 6F polyimides in Figure 2.5........................................ 85 Dianhydride (R) and diamine (Z) elements of polyimides with fluorinated alkoxy side chains................................. 87 Table 2.3. Table 2.4. ........ Table 2.5. Measured and predicted refractive indices of SF5 polyimides .................................................... 88 Table 2.6. Optical properties of PMDA-ODA................................ Table 2.7. Refractive indices of optically transparent polyimides........................ 90 Table 3.1. Color and brittleness of Probimide ® 412 films cured at progressively higher temperatures...................................................... 108 Table 3.2. IR absorption regions of Probimide® 412.................................. 129 Table 4.1. Color and brittleness of Probimide® 32 films cured at progressively higher temperatures......................................................... 144 Table 4.2. IR absorption regions of NMP and xylene........ ......... 89 .............................. 161 Table 4.3. Table 5.1 Partial list of IR absorption bands in Probimide® 32 polyamide-imide.......................... 163 Composition of Merck E7 nematic liquid crystals from Cognard's chromatographic analysis ................................................... 194 Table 5.2. Physical properties of Merck E7 nematic liquid crystals..................... 195 Table 5.3. Settings for EG&G lock-in amplifier used for phasesensitive detection of chopped light................................ 200 List of polymers which were examined as liquid crystal aligning agents and some of their characterisitics..................... 208 Table 5.4. Table 5.5. Relative ratios of the heights of the imide N-C (1370 cm- 1) peaks to the imide C=O (1720 cm - 1) peaks for brushed Probimide 32 layers............................................................................ 216 ACKNOWLEDGMENTS There are so very many people I would like to acknowledge for helping this project go forward, but knowing where to begin is not difficult at all. First, I would like to thank my advisor Professor Peggy Cebe for seeing this project through to the end, and for her ceaseless commitment to my graduate education. Peggy has been an immense source of inspiration to me, as she has been to so many other students. I would also like to thank other individuals whose work has provided direct help in seeing this research go forward. I would like to thank Dr. Enid Sichel, not only for the contributions she has made to the project, but also for her never-ending support and career advice. I would also like to acknowledge UROP student Roland Desrochers for his contributions to the project. I am also grateful to the Minnesota Mining and Manufacturing Company for providing funding for this project. Specifically, I would like to thank Dr. David Cross, Dr. Robert Wenz, and Dr. Timothy Gardner for their consultations and advice. I also thank the various individuals and institutions who donated materials and supplies. First, I would like to thank Dr. Anne St. Clair of the NASA Langley Research Center for providing numerous polyimide samples. I would also like to thank OCG Microelectronic Materials, Inc., Nippon Electric Glass, Amoco Chemical Company, and Nissan Chemical Company. Next, I would like to express my appreciation to the following Professors and their research groups for allowing use of laboratory equipment: Professor Anne Mayes for the spin coater, Professor Michael Rubner for the FTIR, Professor Frederick McGarry for the curing furnace, and Professor Stephen Senturia for the Metricon prism coupler. I would also like to thank the individuals who helped instruct me in using the equipment. Specifically, I would like to thank Dr. Ken Zemach for help with the FTIR, and Dr. Greg Kellogg for help with the spin coater. I also would like to acknowledge my thesis readers Professor Frederick McGarry, Professor David Roylance, and Professor Lionel Kimerling. I immensely appreciate their time and effort. Finally, I would like to thank my colleagues, friends, and family for the love and support they have provided. I thank all recent members of Professor Cebe's Polymer Physics Group for providing a friendly and supportive work environment. I express my wishes for the best of success to all of them. Chapter 1. Introduction and Theory Polyimides are high temperature, high performance polymers with substantial technological importance [1-6]. They exhibit excellent mechanical, thermal, adhesive, and chemical-resistant properties that are highly advantageous for aerospace, electronic packaging, display, waveguide, and other high technology applications. Many polyimides with varying properties and behavior are available commercially. In this thesis, I report the structure and properties of polyimides and examine their use in liquid crystal alignment applications. It is the goal of this chapter to introduce the reader to polyimides, and to introduce some of the main concepts that are essential for understanding the topics addressed in this thesis. First, I discuss in this chapter the synthesis and processing of polyimides. Then, I discuss the key material structure issues that are pertinent to this thesis. Third, I discuss material properties, with my focus on the optical properties of polyimides and nematic liquid crystals. Finally, I discuss the performance of polyimides in nematic liquid crystal alignment applications. 1.1. Synthesis and Processing 1.1.1. General scheme One synthesizes polyimides by a step growth polymerization, followed by a ring closure reaction (see Figure 1.1) [7]. A polycondensation reaction between a dianhydride and a diamine forms a polyamic acid, the precursor to polyimide. This reaction generally procedes in a polar, high boiling point solvent such as y-butyrolactone (GBL, Tb = 2050 C [8]), dimethylacetamide (DMAc, Tb = 166 0C [9]), or N-methylpyrrolidone (NMP, Tb = 2020 C [10]). A high temperature cure, usually 3000C or above, forges a ring closure, or imidization, reaction between the acid and amide groups of the polyamic acid precursor, converting the precursor to polyimide. One can also carry out a chemical imidization at low temperature by adding an acetic anhydride/pyridine solution [7]. Water is removed from the polymer chain during the imidization step. Polyimides usually serve as film coatings. Spin coating, solvent casting, and doctor blading are common processing techniques. Manufacturers also produce thick extruded polyimide films. Subsequent processing of the films depends on the application. 0 11 II C + 0 \ H2N \C /H2N 11 0 Dianhydride H Diamine O H m 0 Polyamic acid (precursor) -H20 O c11 -N \C II O Poy de O Polyimide Figure 1.1. General synthesis scheme of polyimides. PMDA-ODA shown here. 1.1.2. Photosensitive polyimide precursors Photosensitive polyimides [11] which can be patterned by photolithography are of tremendous interest to the electronics industry. Workers at Siemens [12,13] introduced polyimide precursors that could be processed as negative photoresists. The photoprocessing scheme is shown in Figure 1.2. Soluble polyamic esters with photoreactive ester groups crosslink upon exposure to UV radiation. Development of unexposed regions yields an insoluble photopattern of crosslinked polyamic ester precursor. The precursor material requires a high temperature cure after photoprocessing in order to convert it to polyimide. During this step, the side chain crosslinks are released from the main chain. Since the polymer is converted from a soluble polymer (polyamic ester) to an insoluble polymer (polyimide), there is generally a great deal of solvent loss and film shrinkage during this step. In addition, organic molecules which can potentially plasticize the film are released from the polymer during the cure. 1.1.3. Preimidized polyimides Polyimides in solution are generally in their preconverted (polyamic acid) form. That is because the fully converted materials are usually insoluble. Workers at NASA Langley Research Center [14] have investigated techniques of improving polyimide solubility, which include incorporating bridging groups between phenyl rings in the main chain and manipulating the isomeric points of attachment of the bridging groups to the phenyl rings. Some polyimides are highly soluble in their fully converted form and can be coated directly from solution [15,16]. They do not need to be cured for the purpose of imidization, but for the removal of solvent. These fully converted materials are typically soluble in the same organic solvent as polyamic acids, i.e., NMP, GBL, etc. One important class of polyimides are polyamide-imides [17,18], which have the general form: O O It t · 4N 0 c R RN Z C" H 4 n where R and Z are, in general, aromatic groups. Although these polymers are not widely known to be soluble in the fully converted state [19], a fully imidized and soluble polyamide-imide is manufactured by OCG Microelectronic Materials [18]. This material, called Probimide 32, is of substantial importance to the research presented in this document. u II1 ,N• UiR- . -a C CH3 II -uri II O R = CH2 -CH20-C.C6CH2 0 o Polyamic ester (precursor) hv CII" OR - Organics assoc. w/ R Crosslink OR-C4 O Thermal Cure Step Polyimide Figure 1.2. Processing scheme of photosensitive polyimide precursors. 1.1.4. Preimidized, photosensitive polyimides Researchers at Ciba Geigy in 1985 introduced polyimides that are both photosensitive and soluble in the fully converted, imidized state [15,16]. There has been much work done since then investigating their properties [20-27] and photo-reaction mechanism [28-31]. Solubility in the converted state was achieved by the addition of alkyl substituents to the polymer's phenyl groups. These alkyl groups are also involved in the polymer's photochemical reaction. These materials can be processed as negative photoresists from the preimidized polymer in solution without the need for imidization. This material, unlike the photosensitive precursor, undergoes minimal shrinkage and does not contain organic by-products upon process completion. A material of this type, its properties, and its reaction mechanism are discussed in Chapter 3. 1.2. Structure 1.2.1. Macroscopic orientation of polymer films Three levels of macroscopic orientation which might exist in polymer films are depicted in Figure 1.3. Figure 1.3(a) shows a polymer which is an unoriented random coil. This structure is completely isotropic Figure 1.3(b), on the other hand, shows a structure that is in-plane oriented. That is, the molecular chains of the polymer preferentially lie parallel to the plane of the film. This structure is typical of thin coated films [32-42], especially those with rigid backbones. Polymeric films can also orient along one axis of the film, as shown in Figure 1.3(c). One way to induce this type of orientation in thick films is to draw the material [43-49]. In very thin films on the order of hundreds of Angstroms, this type of orientation can be produced by brushing [50-55] the films along one axis with a cloth. The significance of brushed polymer layers in the alignment of liquid crystals will be discussed in section 1.4. 1.2.2. Crystalline packing Though most polyimides are amorphous, some can be semicrystalline [56-66]. Currently, it is difficult to know for certain, prior to synthesis and experimental determination, whether a polyimide with a certain chemical composition will readily crystallize. In addition, since, in general, the softening and melting temperatures of these polymers are extremely high, it is generally difficult to control crystallinity in these materials. However, crystallinity is somewhat controllable in several thermoplastic polyimides [59,60,62-66] designed with low glass transition and melting temperatures. (a) Isotropic (b) In-plane orientation f - I/ • • Irs-~ Uniaxial orientation -~35~L=_ Figure 1.3. Types of orientation in polymer films: a.) isotropic orientation, b.) in-plane orientation, c.) uniaxial orientation. 1.2.3. Chain-chain interactions Many workers believe that charge transfer complexation [67-74], a strong interchain electronic interaction, occurs in polyimides. Dine-Hart and Wright [67], in their early studies of model aromatic imide compounds, first suggested this interaction to take place. The phenomenon is demonstrated in the top portion of Figure 1.4. The lower portion of Figure 1.4 shows the staggered arrangement of adjacent polymer chains discussed by St. Clair [70]. Imide nitrogen donates electrons to their adjacent phenyl groups while the carbonyl groups withdraw electrons from their adjacent phenyl rings. Intermolecular charge transfer complexes result from the opposite partial charges on the aromatic rings of the polymer. Furthermore, it was suggested that this strong interchain interaction is associated with staggered packing of the phenyl rings, with rings of opposite partial charge in maximum contact. dianhydride O diamine II II + 0 Chain 1 + Figure 1.4. Charge transfer interactions in polyimides. Chain 2 1.2.4. Liquid crystalline nematic order and alignment Nematic order Monomeric nematic liquid crystals [75-78] are very important materials for display applications. A typical chemical structure [77,78] of such a material is shown below: B In the above structure, "A" represents a chemical group such as an ester, a phenyl ring, or, perhaps, no group. "B" is typically a hydrocarbon chain 3-7 carbon atoms in length. The nematic liquid crystalline state is a condition in which the molecules in a phase orient along a preferred axis know as the nematic director. The nematic state is depicted in Figure 1.5, where D' is the nematic director. Further discussion of the nematic phase can be found in many texts [75,76]. pD' Figure 1.5. A nematic liquid crystalline phase with director D'. Nematic alignment The difference between nematic order and nematic alignment is a crucial distinction. A slab of nematic liquid crystalline material will form many nematic grains, each with its own nematic director. But without any external influence, the nematic directors of the individual grains will have no preferential orientation with respect to the bulk. This is represented in Figure 1.6 (a). Macroscopic "alignment" [77-81] is the condition in which the nematic directors of the liquid crystals in a slab of material are non-random with respect to the bulk. That is, the liquid crystals are preferentially oriented along a particular axis of the bulk. A "perfectly" or "ideally" aligned specimen will be entirely homogeneous, that is, have only one nematic director. This condition is represented in Figure 1.6(b). The ability to align liquid crystals on the macroscopic level is essential for using of liquid crystals as flat panel display materials. Alignment can be produced, in a general sense, by a number of techniques [77-81]. Such techniques include applying an electromagnetic field, shearing, inducing flow, and treating adjacent surface layers. In this thesis, the treatment of adjacent surface layers to produce nematic liquid crystal alignment is of particular interest, and will be discussed later in this chapter (section 1.4) as well as in Chapters 5 and 6. 1.3. Properties 1.3.1. Quantitative relationships between polymer chemistry and refractive index The index of refraction determines the speed of light in a material through the defining equation: n(c) = v/c (1) where n is the refractive index, co is frequency, v is the speed of light in the material, and c is the speed of light in a vacuum. Refractive index is also a measure of the electronic polarizability of the material. The index of refraction is related to electronic polarizability (a) by the Lorentz-Lorenz equation: [(n2 - 1) / (n2 +2)] (M/p) = (Nav/3co) a (2) where M is molecular weight, p is density, Nay is Avagadro's number, eo is relative permittivity in a vacuum. Refractive index is related to dielectric constant by: (a) \1 ss _o J• s S 14 (b) D' - an axis of the bulk Figure 1.6. Representations of nematic liquid crystalline material. b.) bulk alignment along axis D'. a.) no alignment, and Eo = n2 (3) where e, is the dielectric constant at optical frequency. The refractive index is an important parameter for polyimide applications. For many electronic packaging applications, a low dielectric constant is required for minimizing crosstalk between the copper lines and maximizing signal propagation through the copper itself. The index of refraction is a particularly valuable measure of dielectric properties because it is easily obtained experimentally, even if the material is anisotropic. Polyimide electronic packaging layers have also been suggested to serve as waveguides for optical interconnect technologies [82]. The refractive index is an important parameter in evaluating their performance as waveguides. Researchers have found that fluorinated polymers have low refractive indices. The index of refraction, for example, for polyethylene (-CH 2 CH 2) is 1.51 [83], for poly (vinylidene fluoride) (-CF2CH 2-) is 1.42 [83], and for polytetrafluoroethylene (-CF 2CF 2-) is 1.35 [83]. Groh and Zimmermann [84] predict that the intrinsic lower limit for the refractive index that can be introduced by fluorination is about 1.29. A tremendous amount of work has been done in reducing the refractive indices of polyimides by addition of fluorine to the polymer backbone [85-99]. Because the index of refraction is an important physical property of a material, the relationship between polymer chemical structure and index of refraction is of significant interest to polymer scientist. In addition, since polyimides with low refractive indices are of significant industrial interest, it is important to understand why the index of refraction is reduced. This is an important subject of Chapter 2 of this thesis. Techniques have been developed to predict the index of refraction of a polymer, given its chemical structure. Two methods are discussed below. The first class of techniques discussed here are van Krevelen additive group contribution techniques [100]. The second type of approach is a prediction methodology developed by J. Bicerano [101] which utilizes the concept of atomic bond connectivity [102,103]. Van Krevelen The underlying concept behind van Krevelen [100] group contribution techniques is that polymer properties are "in some way determined by a sum of contributions made by the structural and functional groups in the molecule or repeating unit of the polymer"[100]. The individual contributions, called "increments", that each of these groups makes to a physical property are assumed to be additive. 33 The additive physical property we are concerned with here is the "molar refraction", from which refractive index can be easily calculated. Van Krevelen offers three techniques for refractive index prediction [100]. The first is based on the Lorentz-Lorenz equation, previously shown in equation (2), and shown below in equation (4). [(n2 - 1) / (n2 +2)] (M/p) = (Nav/3eo) aO P RLL (4) P is the molar polarization defined here as equal to the molar refraction RLL. The molar refractions of the individual van Krevelen groups (i) in the backbone are summed according to: RLL = X RLLi (5) It is crucial to note that foreknowledge of the molar volume M/p is required in order to make this calculation. The second technique is based on the equation of Gladstone and Dale [104]: RGD = (n- 1) (M/p) (6) The molar refraction RGD is summed in the same way, using a different tabulated set of group contribution values for the groups that make up the backbone. The molar volume M/p is also required to make this calculation. The third technique does not require experimental values for the density nor any other experimental measurement. This method is based on the equation of Vogel [105]: Rv = nM (7) which states that the Vogel molar refraction RV is simply the product of refractive index (n) and molecular weight (M) in grams/mole. The total Rv is the summation of the individual group contributions. A limitation of this technique is that it does not explicitly take into account the density of the material. Nonetheless, this equation has been shown to give refractive index predictions which are as accurate as those obtained by the more theoretically-based equations of Lorentz-Lorenz and Gladstone-Dale. The Vogel technique is hence a convenient way to estimate refractive index without the need for experimental density data, which is often unavailable. This technique, however, does not factor in the possibility of 34 increased density due to semicrystallinity, chain packing, or other type of ordering. We might therefore expect this method to be more accurate for amorphous polymers with limited chain packing. One limitation of this form of technique, in general, is that contributions of a particular group may differ in different surroundings. For example, the bonding environment around a particular group may differ if attached to an alkyl group as opposed to an aromatic group. However, the technique attempts to correct, as much as possible, for different bonding environments. Another limitation of the van Krevelen method, in general, is that the functional groups are preselected and may not be appropriate or convenient for complicated novel polymers. A reasonable approach to assembling an appropriate set of contributing groups must be employed, and sometimes best guesses must be made. For example, correction factors for several ring structures have been calculated by van Krevelen, although no value is specifically available for the imide ring structure present in polyimides. Since most ring structures have very close correction factors, the value for a similar ring, a piperidyl ring, is used instead in Chapter 2. A sample van Krevelen/Vogel calculation for polystyrene is shown in Figure 1.7. Bicerano A new method for polymer property prediction developed by Jozef Bicerano of Dow Chemical Company has recently become available [101]. This is not a group contribution method, but a topological method that takes into account all of the atoms and bonds present in the material. It utilizes four connectivity indices [102,103] and other parameters which are calculated as described below and demonstrated for polystyrene in Figure 1.8. Because there is no preselection of chemical contribution groups, we should expect this technique to be especially useful for complex polymers such as polyimides. The hydrogen suppressed repeat unit is drawn (with a bond drawn at only one end of the repeat unit connecting the unit to the next one). At the location of each atom, the number of non-hydrogen atoms to which it is connected is written. The value at each atom represents the zeroeth order simple connectivity index 8 of the atom. The zeroeth order simple (atomic) connectivity index for the entire molecule OX is then the sum of the negative one-half powers of the connectivities of all its atoms: ox = 1 (1/I) ( 8) Van Krevelen Calculation (Polystyrene) 20.64 CH- (ar) 21.4 --0 123.51 Rv = 165.55 M = 104 g/mol n = R, /M = 1.592 nexp = 1.591 Figure 1.7. Sample calculation of the index of refraction of polystyrene using the van Krevelen / Vogel technique. The predicted value, 1.591, is in excellent agreement with the experimental value 1.592 [83]. The product of the atomic connectivities of every two adjacent atoms is then written at the bond line between the two atoms. The value at each bond line represents its bond index p. The first order simple (bond) connectivity index for the entire molecule is then calculated as shown in equation (9). This index, though used to predict other properties, is not present in Bicerano's refractive index correlation. 'X = I (1/•p) (9) Next, on a new hydrogen suppressed repeat unit, zeroeth order valence connectivity indices 8•, given by equation (10), for each atom, are written at the location of each atom. 8v = (Zv - NH) / (Z - Zv -1) (10) In equation (10), Zv is the number of valence electrons, NH is the number of adjacent hydrogen atoms, and Z is the atomic number. For second row atoms carbon, nitrogen, oxygen, and fluorine, equation (10) reduces to 8v = Z v - NH, that is, the atom's valence minus the number of hydrogen atoms attached. First order valence connectivity indices for the bonds of the molecule are calculated as the products of the adjacent zeroeth order valence indices and are written at the bond lines. Atomic and bond valence connectivity indices for the entire atom are calculated as shown in equations (11) and (12). OXV = e (1/~8v) (11) 1xv = I (1/4V) (12) In the development of correlations for property prediction [101], Bicerano uses these connectivity parameters, as well as additional correction factors which account for the overestimation or underestimation of the contribution of certain groups or atoms due to, for example, orbital conjugation, hydrogen bonding, or unusually low polarizability. Bicerano's correlation for predicting refractive index at room temperature is shown in equation (13). n = 1.8853 + 0.024558 (17 OXV - 20 OX -121v. - 9 Nrot + Nref) /N (13) In equation (13), N is the total number of non-hydrogen atoms. Nrot is the number of rotatable, non-aromatic single bonds in the repeat unit that change the spacial coordinates of one or more atoms upon rotation. Bonds in "floppy" rings contribute one-half to Nrot. Nref is a correction factor given by equation (14). Nref = -11NF - 3NCI-Ph + 18NS + 9 Nfused + 12NHB + 32NSi-si (14) In equation (14), NF is the number of fluorine atoms, NS is the number of sulfur atoms, NCI-Ph is the number of chlorine atoms bonded to phenyl rings, Nfused is the number of rings which are present in attached-edge multi-ring (e.g., aromatic imide) structures, NHB is the number of hydrogen bonding moieties, and NSi-si is the number of delocalized and conjugated silicon-silicon bonds. Only Nfused and NF are relevant to the polyimides discussed in this work. Using 183 polymers with refractive indices ranging from 1.34 to 1.69, a correlation coefficient of 0.977 and a standard deviation of 0.016 was obtained by Bicerano for equation (10). The standard deviation represents less than 1% of the average refractive index of the data set. Thus, we should expect this method to be highly accurate. 1.3.2. Property changes associated with charge transfer complexation Charge transfer complexation (see section 1.2.3) is believed to have a substantial effect on the properties of polyimides [67-74,85-99]. Examples of property changes believed to be associated with charge transfer complexation include increased yellow coloration, glass transition temperature, dielectric constant, refractive index, and optical attenuation. Workers have found that charge transfer complexation is mitigated when separator groups, chain kinks, and bulky chemical substituents are introduced into the polyimide backbone [72-74,85-99]. Introduction of these chemical groups have marked effects on properties. For example, St. Clair et al. [73,74,89,96] have found that incorporation of charge transfer-reducing groups reduce coloration and dielectric constant. Fryd [72] found that addition of such groups reduce glass transition temperature. Feger et al. [98] have found that addition of bulky fluorinated groups reduces optical losses in waveguides. Bicerano Calculation (Polystyrene) 6 Repeat unit Simple Connectivity Valence Connectivity 8v = valence minus # adjacent H 8 = # adjacent non-H atoms pijv = 8V x 8j •j = 68 x 8j N=8 6 o = 1 (1/48) [atomic connectivity index Nrot =-3 Nref - I X= (1/ 3) bond = (6/N2) + (2/N3) = connectivity index o 5.397 = (1/42) + (6/q3) + (1/N4) = 1 = (4/N4) + (4/46) + (1/49) = 3.966 4.671 X Y = (2/q6) + (4/49) + (3/N12) = 3.016 Nref = -11NF - 3 NCI-Ph + 18Ns + 9 Nfused + 12 NHB + 32 Nsi.si n = 1.885312 + 0.024558 x (17 OXv - 20 OX -12 IXV - 9 Nrot + Nref) / N n (predicted) = 1.604 n (experimental) = 1.592 Figure 1.8. Bicerano prediction for the refractive index of polystyrene. The estimated value, 1.604, is within 0.75% of the experimental value, 1.592 [83]. 1.3.3. Anisotropy of optical properties Birefringence is a term which is defined as a difference in refractive index in a material along different spacial axes. However, it is very important that the term not be used loosely. That is, we must be precise in how we define the axes of comparison. In Figure 1.9(a-e), we review a number of ways in which birefringence can be defined. Examples of optical anisotropy on three different levels are presented: molecular, microscopic, and macroscopic. Molecular birefringence Optical anisotropy on the molecular level is shown in Figure 1.9(a). In the molecule in this figure, the refractive index along the chain axis n// is different from the refractive index perpendicular to the chain axis n±. One might define the molecular birefringence Anmol as: Anmol = n// - n± (15) In this particular example, n// should be greater than nj. This is because the polarizability of the aromatic ring is greater in its plane than normal to it. The axis parallel to the main chain must always be in the plane of the phenyl rings of the polymer, while axes perpendicular to the main chain will, in general, not be. For a molecule in which the highly polarizable aromatic units are perpendicular to the main chain, e.g., polystyrene, n// will be less than n± [33,34]. Microscopic birefringence Birefringence on the microscopic level is shown in Figure 1.9(b,c). A polymer spherulite appears in Figure 1.9(b) in which the refractive index along the radial axis nr is different from the refractive index along the tangential axis nt. It might be convenient for the case in Figure 9(b) to define birefringence of the spherulite Ansph as: Ansph = nr - nt (16) The relationship of nr to nt will depend upon the orientation of the main chain with respect to the radial axis, as well as the relationship between n// and nl described above. Figure 1.9(c) shows a nematic liquid crystalline grain in which the refractive index along the extraordinary axis ne is different from the refractive index along the ordinary axis no. This is another form of microscopic birefringence. We might define birefringence of the liquid crystalline phase AnLC as: AnLC = ne - no (17) We can expect ne to be, most often, greater than no because the molecules of organic liquid crystals typically have their primary polarizable groups along the long axis of the molecule. Macroscopic birefringence Birefringence on the macroscopic level is shown in Figure 1.9(d,e). An in-plane oriented polymer film is shown in Figure 1.9(d) in which the bulk in-plane index nTE is different from the bulk out-of-plane index nTM. This film structure was previously shown in Figure 1.3(b). The difference between nTE and nTM will depend upon whether the main chain is more polarizable along its main chain or normal to its main chain. If, for example, the molecule is more polarizable along its main chain (e.g., polyimide), nTE will be greater than nTM, as the bulk film consequently will be more polarizable in the plane of the film. On the other hand, if the polymer is more polarizable normal to the main chain (e.g., polystyrene), nTM will be greater than nTE [33,34]. Here, we can define the birefringence of the film Anfilm as: Anfilm = nTE - nTM (18) An aligned liquid crystal film is shown in Figure 1.9(e) with differing refractive indices nTE,//, nTE,I, and nTM. The condition of macroscopic alignment was previously shown in Figure 1.6(b). In fhe case that the liquid crystals are perfectly aligned, nTE,// should be equal to no in Figure 1.9(c), while nTE,± and nTM should be equal to ne from Figure 1.9(c). polymer molecule -0CD (b) nt n polymer spherulite Figure 1.9. Examples of optical anisotropy. a.) birefringence on the molecular level, b.) birefringence on the microscopic level in a polymer spherulite, (c) nematic liquid crystalline domain no (d) n TM in-plane oriented polymer film nTE riTE TM aligned nematic liquid crystaline film nTE, S TE,// Figure 1.9. continued. Examples of optical anisotropy. c.) birefringence on the molecular level in a nematic grain, d.) bulk birefringence in an in-plane oriented film, e.) bulk birefringence in a film of aligned liquid crystals. For the general case of a macroscopic birefringence, we can imagine a block of material in Cartesian axes as shown in Figure 1.10. We can define three refractive indices nx, ny, and nz, as well as three macroscopic birefringences as shown in equations (19), (20), and (21): Anxy = nx - ny (19) Anxz = nx - nz (20) Anyz = ny - nz = Anxz - Anxy (21) We can also define an average refractive index as: navg = (1/3) (nx + ny + nz) (22) For an in-plane oriented polymer film which is isotropic in the plane of the film, as shown in Figures 1.3(b) and 1.9(d), average refractive index is: navg = (2 nTE + nTM) / 3 (23) This focus of this thesis is macroscopic optical properties. However, others [106] have done extensive investigations of the relationships between intrinsic molecular birefringence, crystalline fraction, crystalline orientation function, and macroscopic birefringence. Molecular vs. microscopic vs. macroscopic birefringence It is important to realize that a material can be birefringent on either the molecular, microscopic, or macroscopic level, yet not be birefringent on any of the other two. For example, most polymers have refractive indices that are different along the main chain as compared to normal to the main chain. However, a film or block of such a polymer will not necessarily be birefringent on the macroscopic or even the microscopic level. It may be entirely amorphous and unoriented. Likewise, a material can be birefringent on the microscopic level, but not on the macroscopic level. Figure 1.10 shows a block of polymer which is semicrystalline. The crystallites are perfectly spherical. Assuming that, for the spherulites, nr • nt, the material will be highly birefringent on the microscopic level. When placed between crossed polarizers, a significant quantity of light will be transmitted. However, if one measures the bulk refractive indices along axes x, y, and z of the block, one will find that they are exactly the same! Between crossed polars, rotation of the block about the normal axis will not change the transmitted intensity. It is does not matter whether, for example, the x-axis is 00 or 450 to the polarizer. That is, we have the situation in which the material is highly birefringent (microscopically), though the bulk refractive indices are identical in every direction. The situation described above may occur for any situation in which there is microscopic birefringence, but no orientation with respect to the bulk. A slab of unaligned nematic liquid crystals, for example, will be "birefringent" locally, but "non-birefringent" if we define the optical axes with respect to the bulk sample. A slab of aligned liquid crystals, on the other hand, will be birefringent on both the microscopic and macroscopic levels. perfectly spherical crystallites 0 m polymer block /I 0000 03 Uh I I,. Figure 1.10. A thick semicrystalline polymer block that is microscopically birefringent but has nx=ny=nz. 1.3.4. Measurement of index of refraction There are many techniques in existence that can be used to measure the refractive index of a material [107-110]. These include ellipsometry, Abbe refractometry, and fluid matching. Another technique of recent interest is called waveguide mode prism coupling [108,109], and is discussed here. The application of the prism coupling technique is shown in Figure 1.11. The film is pressed to the base of a high index prism. A polarized laser is focused on one face of the prism. The light refracts into the prism, totally internally reflects at the base of the prism, and is detected as it leaves the opposite face. When the angle of incidence e is equal to that of an allowed mode in the film, the light propagates into the film and there is a sharp drop in light intensity at the detector. Also, when O drops below the critical angle of the polymer film, the laser refracts into the film, and there is a large permanent drop in detected intensity. The refractive index of the material can be calculated from either the waveguide mode "coupling" angles or from the critical angle. The critical angle approach, while slightly less accurate than waveguide propagation method, is preferable to use in thick films in which waveguide modes are poorly pronounced. Workers at Bell Labs developed the prism coupling method in the early 1970s [108]. Today, a commercial instrument, the Metricon prism coupler, applies this technique. Using the Metricon instrument, one can measure both the in-plane index (nTE) and out-of-plane index (nTM) of a thin film by changing the polarization of the incident light. The Metricon prism coupler is equipped with both waveguide mode and critical angle measurement capability. Many workers have used the prism technique to measure the refractive index and optical anisotropy of polyimide and other polymer films [32-41,43]. FILM Figure 1.11. Prism coupling technique for measuring index of refraction. 1.3.5. Optical transmission model for liquid crystal alignment This section is dedicated to the discussion of an adequate framework for evaluating the alignment of nematic liquid crystals in a film. Nematic liquid crystals are birefringent on a microscopic level. The molecules in each liquid crystalline domain are oriented preferentially along a director with the index of refraction greater along the ordinary axis than along the extraordinary axis. The microscopic birefringence of the liquid crystalline phase is a function of the material itself and is independent of any external forces which may act to align the liquid crystals with respect to the bulk. However, in this section, we are concerned with evaluating the macroscopic alignment of the liquid crystals, that is, how well they are oriented with respect to a given axis of the bulk. Optical model for an aligned liquid crystal plate Let D' be the nematic director of an aligned liquid crystal plate (such as that found in Figure 1.9(e)), with D" the ordinary axis orthogonal to D' and in the plane of the plate. Also, let OP be the axis of the polarizer, and OA be the axis of the analyzer. Let angle X be the angle between the polarizer and analyzer, angle 0 be the angle between the polarizer and nematic vector. These vector are shown in Figure 1.12 below. Figure 1.12. Graphical representation of a parallel-aligned liquid crystal plate between a polarizer (P) and analyzer (A). An electric field plane wave passing through the polarizer enters the liquid crystal plate. When D' is not parallel or perpendicular to OP, the electric field vector is divided into two components that have mutually orthogonal displacement vectors but differing velocities of propagation. The two waves emerge from the plate with a phase difference 5 given by [111]: (24) 8 = (2n / X)(An) h where: , = wavelength of light in free space An = difference between ordinary and extraordinary refractive indices h = plate thickness The intensity I of light passing through the crystal plate, polarizer, and analyzer is given by [111]: I = E2 {cos 2 x - sin 20 sin 2(4-X) sin2 (8/2)) (25) When the polarizer and analyzer are crossed, i.e. X= nt/2, intensity II is given by: II = E2 sin2 24 sin 2 (8/2) (26) When the nematic vector D' is aligned parallel or perpendicular to the polarizer axis OP, the value of Ij is identically zero. That is, according to equation (26): II = Ij, min = 0, when 4 = 0, 7/2, in, ... (27) On the other hand, the angular position of the liquid crystal nematic vector which produces maximum intensity II, max is 450 with respect to the polarizer axis. The maximum intensity in equation (26) is given by: II = Ij, max = E2 sin2(8/2), when 4 = n/4, 37r/4, 57t/4, ... (28) If we normalize the equation for I1() given by equation (26) by the maximum intensity II, max given by equation (28), we get the following simple expression for I(O)/I(f/4), that is, the intensity at angle 4 normalized by the maximum intensity found at S----n/4: I(0) / I (c/4) = sin 2 (20) (29) This relationship is plotted below in Figure 1.13. 0 in l,~I 10 >1 C .O1 .01 .001 .0001 0 45 90 135 Angle (degrees) 180 Figure 1.13. Plot of I(0)/I(450 ) vs. 0 from equation (29). Elliptically polarized light Figure 1.14 shows the most general case of elliptically polarized light [107,112,113]. The direction of polarization systematically changes in an elliptical pattern as it propagates. The polarization state of the electric field can be described by three parameters, which are represented in Figure 1.14. The first parameter is the handedness, right (clockwise) or left (counterclockwise). The second parameter is the azimuth a, that is, the angle between the semimajor axis of the ellipse and the defined reference axis. The third parameter is the ellipticity 0, which is the ratio between the lengths of the semimajor and semiminor axes of the ellipse, or b/a in Figure 1.14. b left handed a right handed Figure 1.14. Elliptically polarized light with handedness (left or right), azimuth (a), and ellipticity (P = a / b). Linearly and circularly polarized light are special cases of elliptically polarized light. Linearly polarized light is the case where 3 = 0. It can possess any value of a, but has no handedness. Circularly polarized light, on the other hand, is the case where 101 = 1. It can be left or right handed, but has no azimuth. Misalignment In the case of a perfectly aligned specimen, polarized light passing through the specimen will move from an original state of linear polarization (a = 0 and 1 = 0) to another state of elliptical polarization depending upon the thickness of the film and the orientation with respect to the polarizer. This is true except for the case where the liquid crystals are aligned 0 or 900 with respect to the polarizer and in which there is no change in the alignment state. Now consider a film of liquid crystals in which the liquid crystals are not in a perfect state of alignment. A case in which there are regions of liquid crystals misoriented with respect to the bulk alignment axis is depicted below in Figure 1.15. Direction of alignment Figure 1.15. Parallel-aligned liquid crystals with inhomogeneities. Consider only the case in which the polarizer and analyzer are crossed and the liquid crystal alignment director is parallel to the polarizer. We showed previously that under these conditions in an ideally aligned system, no light is transmitted, i.e., intensity is zero. However, if the system has misaligned liquid crystal domains, some light can be transmitted as a consequence of random elliptical polarization produced as light travels through the unaligned regions. Thus, unlike the ideal case, transmitted intensity will be greater than zero. Each distinct misaligned nematic domain will have its own orientation with respect to the polarizer. The following exercize will clarify this concept. Imagine a linearly polarized laser beam in azimuth ao passing through a LC specimen with random inhomogeneities "a", "b", "c", "d", and "e". This is shown in Figure 1.16. As the laser penetrates the liquid crystalline sample at point "1", the light passes through inhomogenous phases "a" and "b". Both "a" and "b" have unique nematic directors and unique optical path lengths. Light incident at point "2" of the specimen, however, travels through inhomogenous phases "c", "d", and "e", which have their own unique directors and optical path lengths. The light emerging at points 1' and 2' will have unique ellipticities 3 1 and 12 and unique azimuths al and a2 different from the initial polarization. A large beam of light, hence, will emerge from the sample with a distribution of polarization states depending upon the significance of the misalignment. Consequently, some light will not be blocked by the analyzer. Furthermore, since there is no specific relationship between P1and 12 or between al and a2, there is no compensator, e.g., a 1/4 wave plate, that can single-handedly negate the ellipticity of the emanating light. P=E0 a:=czo LASER: --a 1 • I V zy I 12 w| m 21 02, (X2 01, al I1 •2 •0 al # a2 # 0X Figure 1.16. Rays of light passing through different points 1 or 2 of the liquid crystal sample and emerging at 1' or 2' with different polarization states. Domains a-e represent regions of differing optical retardation. Characterizing the Misaligned State Let us suppose that a liquid crystal film is placed between a crossed polarizer and analyzer with the alignment axis at an angle 0 with respect to the polarizer. Let us now define a liquid crystal alignment parameter "A", which is given by the ratio of the =00 to the experimentally measured experimentally measured transmitted intensity at -- transmitted intensity at 0=450. The parameter "A" is, essentially, the ratio of minimum to maximum intensity. A - I(00) / I(450) (30) The meanings of the values of A are shown in Table 1.1. It has already been shown in equation (29) that the ideal value of A for a perfectly aligned liquid crystal film is zero. From the previous discussion of misalignment, it is clear that for a film with misalignment, A will be greater than zero. For the case that there is no alignment, A is unity, as there is no distinguishable difference between the 00 and 450 conditions. Hence, A will, in general, be a parameter between zero and one inclusively, with low values representing strong alignment, high values representing weak alignment, zero representing perfect alignment, and one representing no alignment. Note that a value of A greater than one would represent the case in which the material is aligned with a greater preference along the 450 axis than along the alignment director that was initially assumed. Table 1.1. Meaning of the ratio of transmitted intensity at 00 (MIN) to transmitted intensity at 450 (MAX). MIN/MAX (A) ratio Meaning A=0 LCs perfectly aligned 0 <A < 1 LCs have intermediate level of alignment e.g., A is small, but > 0 LCs strongly aligned A is large, but < 1 LC weakly aligned A= 1 LCs not aligned A> 1I LCs aligned with preference to 450 axis Using equations (29) and (30), we can also define an "effective angle of misalignment" O: A = sin 2 (20) (31) The parameter 0 represents the angle at which the liquid crystals appear to be misaligned with respect to the alignment (00) axis. For the case that A=0, i.e., perfect alignment, 0--=0. For the case that A=1, i.e., no alignment, 0=450. The 0-=450 condition, means that the liquid crystal nematic phases have their directors oriented along any axis with equal probability. Equation (31) is not valid for the odd case that A>1. 1.4. Performance: Liquid Crystal Alignment Layers Polyimides are important as electronic packaging materials, as engineering plastics for high technology applications, and as alignment films for liquid crystals. Much of this thesis focuses on their use as liquid crystal alignment films. A discussion of the alignment of liquid crystals on polymer surfaces follows. 1.4.1. Liquid crystal surface alignment on brushed polymer films The surface alignment of liquid crystals is a crucial element in the performance of LCD technology [77-81]. "Alignment layers" at surfaces of liquid crystal films, as shown in Figure 1.17, define the axis of macroscopic liquid crystal orientation at that surface. Alignment layers are typically rubbed polyimides or obliquely evaporated SiOx thin films. In the case of rubbed polyimides, it has been observed that liquid crystals that come in contact with a rubbed polymer film align in the direction the polymer is rubbed. NEMATIC LIQUID CRYSTALS S - -. - --- -F B BRUSHED POLYMER ALIGNMENT FILM AXIS OF BRUSHING Figure 1.17. Alignment of liquid crystals on brushed polymer surfaces. The liquid crystals can order in the direction of the rubbing at an inclination angle with respect to the horizontal surface. This is known as the angle of surface pretilt [114,115]. Depending upon the particular device technology, a specific pretilt angle can be desired for optimal switching performance. The liquid crystal pretilt angle is a function of the polymer alignment material, the nematic liquid crystal, and the way in which the alignment films are processed [116-119]. The generation and control of pretilt was not, however, investigated in this thesis. The brushing process is accomplished industrially on a spinning wheel covered with a silk, rayon, or other cloth. This is shown in Figure 1.18. On the laboratory scale, however, films can be brushed by hand to create liquid crystal alignment, and complex apparatus is not required. BUFFING WHEEL I POLYMER SURFACE Figure 1.18. A polymer surface being rubbed by a brushing wheel. Although rubbed polyimides perform as excellent liquid crystal alignment layers, the mechanism of alignment is not well understood. Brushing-induced surface alignment is often referred to as a "black art", as it has both baffled and intrigued scientists for decades. Berreman and others [120-122] have postulated that liquid crystals align along surface grooves created by the rubbing process. More recently, workers have shown that liquid crystal alignment is more likely associated with the orientation of chemical groups in the alignment polymer [48-55]. Geary et al. [50] observed that polymers that are oriented and highly crystalline are effective liquid crystal alignment films. As to why liquid crystals would align on oriented surface is a matter of speculation, though it has been suggested [123] that intermolecular interactions between the polymer and the liquid crystal are involved. 1.4.2. Alternative treatments to polymer layers for liquid crystal alignment There are a number of serious disadvantages of the brushing process as a means of inducing alignment of liquid crystals. First, the process can generate particulates which can affect the quality of display devices. Second, the buffing can create static charge, which can interfere with the device electronics. Third, the brushing can only be performed on planar surfaces. Notched or curved substrates can present difficulties in achieving brushing-induced alignment. Fourth, brushing is a costly and energy consuming mechanical process. Finally, brushing is a surface-uniform technique which cannot be used directly to pattern a device. A number of alternative, non-contact liquid crystal surface alignment processes have been proposed and demonstrated. For example, workers have investigated the noncontact formation of surface grooves as a means of producing alignment [124,125]. Also, Langmuir-Blodgett films have also been used to align liquid crystals [126-128]. Recently, Schadt and others [129-131] have demonstrated the alignment of liquid crystals on films of a photosensitive cinnamate polymer (polyvinylmethoxycinnamate) [132] that have been exposed to linearly polarized ultraviolet light (LP-UV). Schadt [129] proposes that an anisotropic structure, as shown in Figure 1.19, is forged in the polymer alignment film during the exposure to LP-UV. The top portion of Figure 1.19 shows the chemical groups which react, while the bottom portion depicts the proposed oriented structure. The polymer, which crosslinks upon exposure to UV, is believed to crosslink in a preferential way upon exposure to LP-UV which results in an anisotropic distribution of crosslinked side chain molecules, exerts an orienting force upon the main chain of the polymer, and depletes uncrosslinked side chains oriented along the LP-UV axis. The authors believe that the induced orientation in the polymer is associated with the liquid crystal alignment, which was found to be perpendicular to the axis of the LP-UV. Other researchers have investigated linearly polarized light as a way producing alignment in liquid crystals. One might refer to this class of techniques as "photoalignment". Ichimura et al. [133-135] have investigated the alignment of liquid crystals on azobenzene films that have been exposed to linearly polarized UV. The azobenzene dyes undergo a 'trans-cis' photoisomerization upon exposure to the linearly polarized light [136,137]. Gibbons et al. [138-142] have investigated the optical control of liquid crystal alignment using azo dye doped polyimide alignment films and exposing liquid crystal cells to linear polarized laser beams. Gibbons' group is interested in photo-orientation processes not only for display applications, but also for holography, optical data storage, and other potential applications of liquid crystals which require their macroscopic alignment. o0 LP-UV axis .o LP-UV axis Figure 1.19. The photochemical reaction in polyvinylmethoxycinnamate and the oriented structure, proposed by Schadt [129], that is formed by exposure to linearly polarized UV. Hasegawa and Taira [143] demonstrated the alignment of nematic liquid crystals on films of an aliphatic/aromatic polyimide that have been exposed to linearly polarized deep ultraviolet light. They found alignment to occur, however, in only a tiny window of exposure time, below and above which, no alignment was seen. The authors suggest that the anisotropic depolymerization of the polyimide film is responsible for the alignment. They argue that at short exposures, the polymer begins to degrade preferentially along the LP-UV axis, as shown in Figure 1.20. This process results in an oriented structure in the polymer film that is preferential to the axis in the plane of the film perpendicular to the LPUV. At the exposure dosage critical for producing alignment, the polymer film is in a state of optimal orientation. At this point, chains parallel to the LP-UV become depleted and further exposure to LP-UV destroys chains perpendicular to the UV axis as well, thus, destroying the alignment. LP-UV AXIS ! 1400, K w ORIENTATION AXIS Figure 1.20. Anisotropic chain degradation proposed to occur [143] in a polyimide film during exposure to linearly polarized UV. There is precedent for the belief that deep UV depolymerizes the polyimide. For example, Kambe [144] discusses the degradation of mechanical properties of aromatic polyimide, polyamide, and polyamide-imide films that have undergone very long exposures to deep UV. In addition, polyimides are well known to be destroyed during deep UV laser ablation processes [145-149]. In analogy, the LP-UV here might be thought of as a light, anisotropic surface ablation. 1.4.3. An LCD device: the TN-LCD Liquid crystals displays are devices which take advantage of the optical properties of macroscopically oriented liquid crystalline materials in order to produce switchable pixels. So that it is clear why surface alignment is important in the display industry, an example of a simple liquid crystal device, known as the TN-LCD, or "twisted nematic liquid crystal display" [150], is briefly explained here, and also shown in Figure 1.21. It should be noted, however, that there are other types of LCDs [151], as well as other potential applications of liquid crystals. A TN-LCD cell consists of a nematic liquid crystal phase which is gradually rotated by 900. The 900 rotation is achieved by inducing surface-alignment directors at each liquid crystal-polymer interface that are perpendicular to each other. The twisted liquid crystal film is sandwiched between two substrates, each coated with the appropriately treated alignment film. Each substrate contains a polarizer with the axis of polarization parallel to alignment director defined by the adjacent alignment film. There are also electrodes atop each substrate so that an electric field can be applied to the cell. In the "off" state, the twisted TN layer rotates the direction of polarization of the entering light by 900. The second polarizer prevents light from passing through, and the cell appears dark. However, when voltage is applied to the two electrodes on each surface of the TN layer, the liquid crystals reorient parallel to the applied field. In that state, the layer no longer twists the direction of polarization of the light passing through, and the cell appears bright. When the electric field is removed, the liquid crystals return to their default alignment state, which is the twisted nematic state. Also note that if the two polarizers in the device were perpendicular instead of parallel, the cell would function in reverse, i.e., as off/on instead of on/off. Light cannot pass ayer 4 Light passes through Polarizer Substrate Electrodes ,coated with alignment layer Substrate Polarizer state: (b) On voltage applied Light Figure 1.21. States of a twisted nematic liquid crystal display (TN-LCD) device. a.) on state and b.) off state. 1.5. References 1. K.L. Mittal, Ed., Polyimides: Synthesis, Characterization,and Applications, Plenum Press, New York (1984), Proceedings of the First Technical Conference on Polyimides, Ellenville, NY (1982). 2. W.D. Weber and M.R. Gupta, Eds., Recent Advances in Polyimide Science and Technology, SPE, Inc., Poughkeepsie, NY (1987), Proc. of the Second Tech. Conference on Polyimides, Ellenville, NY (1985). 3. C. Feger, M.M. Khojasteh, and J.E. McGrath, Eds., Polyimides: Materials, Chemistry, and Characterization,Elsevier, Amsterdam (1989), Proceedings of the Third Technical Conference on Polyimides, Ellenville, NY (1988). 4. D.Wilson, H. Stenzenberger, and P. Hergenrother, Eds., Polyimides, Chapman and Hall, NY, (1990). 5. S.D. Senturia, in Polymersfor High Technology - Electronics and Photonics, M.J. Bowden and S.R. Turner, Eds., ACS, Washington, D.C., p 428 (1987), Symposium of the ACS Div. PMSE at the 192nd meeting of the ACS, Anaheim, CA (1986). 6. J.W. Verbicky, Jr., "Polyimides" in Encylopedia of Polymer Science and Engineering, John Wiley and Sons, Inc., New York (1987). 7. F.W. Harris, in Polyimides, p.1, D.Wilson, H. Stenzenberger, and P. Hergenrother, Eds., Chapman and Hall, NY, (1990). 8. Amoco Chemical Company material safety data sheet, y-butyrolactone (1993). 9. Baxter Healthcare Corporation material safety data sheet, dimethyl acetamide (1989). 10. Mallinckrodt, Inc., material safety data sheet, N-methyl pyrrolidone (1985). 11. H. Hiramoto, Mat. Res. Soc. Symp. Proc., 167, 87 (1990). 12. R. Rubner, H. Ahne, E. Kuhn, and G. Kolodziej, Photogr. Sci. Eng., 23(5), 303 (1979). 13. H. Ahne, H. Kruger, E. Pammer, and R. Rubner, in Polyimides: Synthesis, Characterization,and Applications, p. 905, K.L. Mittal, Ed., Plenum Press, New York (1984). 14. T.L. St. Clair, A.K. St. Clair, and E.N. Smith, in Structure-Solubility Relationships in Polymers, p.199, F.W. Harris and R.B. Seymour, Eds., Academic Press, New York (1977). 15. J. Pfeifer and 0. Rhode, in Recent Advances in Polyimide Science and Technology, p.336, W.D. Weber and M.R. Gupta, Eds., Society of Plastic Engineers, New York (1987). 16. 0. Rhode, P. Smolka, P. Falcigno, and J. Pfeifer, Polym. Eng. Sci., 32(21), 1623 (1992). 17. H.R. Slater, Advanced Materials and Processes, 1, 19 (1995). 18. OCG Microelectronic Materials, Inc., product information, Probimide 32 polyamideimide resin. 19. Personal communication. Dr. Doug Fjare, Amoco Chemical Company, and Dr. Brian Auman, DuPont Corporation. 20. K.K. Chakravorty and C.P. Chien, SPIE Int. Conf. on Adv. in Intercon. and Packaging, 1389, 559 (1990). 21. C.P. Chien and K.K. Chakravorty, SPIE Opical. Thin Films III: New Developments, 1323, 338 (1990). 22. R. Mo, T. Maw, A. Roza, K. Stefanisko, and R. Hopla, Proc. 7th Int. IEEE VLSI Multilevel Intercon. Conf., 390 (1990). 23. T. Maw and R. Hopla, MRS Symp. Proc., 203, 71 (1991). 24. T. Maw and R. Hopla, MRS Symp. Proc., 227, 205 (1991). 25. T. Maw, M. Masola, and R. Hopla, Polym. Mat. Sci. Eng., 66, 247 (1992). 26. M. Ree and K. Chen, Polym. Prep., 31(2), 594 (1990). 27. M. Ree, K. Chen, and G. Czornyj, Polym. Eng. Sci., 32(14), 924 (1992). 28. H. Higuchi, T. Yamashita, K. Horie, and I. Mita, Chem. Mater., 3, 188 (1991). 29. J. Scaiano, A. Becknell, and R. Small, J. ofPhotochem. Photobio.A, 44, 99 (1988). 30. A. Lin, V. Sastri, G. Tesoro, A. Reiser, and R. Eachus, Macromol., 21(4), 1165 (1988). 31. J. Scaiano, J. Netto-Ferreira, A. Becknell, and R. Small, Polym. Eng. Sci, 29(14), 942 (1989). 32. T.P. Sosnowski and H.P. Weber, Appl. Phys. Lett., 21(7), 310 (1972). 33. W.M. Prest and D.J. Luca, J. Appl. Phys., 50(10), 6067 (1979). 34. W.M. Prest and D.J. Luca, J. Appl. Phys., 51(10), 5170 (1980). 35. T.P. Russell, H. Gugger, and J.D. Swallen, J. Polym. Sci: Polym. Phys., 21, 1745 (1983). 36. S. Herminghaus, D. Boese, D.Y. Yoon, and B.A. Smith, Appl. Phys. Lett., 59(9), 1043 (1991). 37. D. Boese, H. Lee, D.Y. Yoon, J.D. Swallen, and J.F. Rabolt, J. Polym. Sci.: Polym. Phys., 30, 1321 (1992). 38. S.C. Noe, Doctoral Thesis, Massachusetts Institute of Technology (1992). 39. S.C. Noe, J.Y. Pan, and S.D. Senturia, Polym. Eng. Sci., 32(15), 1015 (1992). 40. R. Burzynski and P. Prasad, Polymer, 31(4), 627 (1990). 41. S. Bai, R. Spry, D. Zelmon, U. Ramabadran, and J. Jackson, J. Polym. Sci.: Pt. B, Polym. Phys., 30, 1507 (1992). 42. J. Machell, J. Greener, and B. Contestable, Macromol., 23, 186 (1990). 43. K. Nakagawa, J. Appl. Polym. Sci., 41, 2049 (1990). 44. V.E. Smirnova and M.I. Bessonov, in Polyimides: Materials, Chemistry, and Characterization,p.563, C. Feger, M.M. Khojasteh, and J.E. McGrath, Eds., Elsevier, Amsterdam (1989). 45. T. Kinugi, I. Aoki, M. Hashimoto, Kobunshi Ronbunshyu, 38, 301 (1981). 46. Y. Aihara and P. Cebe, Polym. Eng. Sci., 34(16), 1275 (1994). 47. J. Teverovsky, D. Rich, Y. Aihara, and P. Cebe, J. Appl. Polym. Sci., 54, 497 (1994). 48. H. Aoyama, Y. Yamazaki, N. Matsuura, H. Mada, and S. Kobayashi, Mol. Cryst. Liq. Cryst., 72, 127 (1981). 49. N. van Aerle and J. Tol, Macromol., 27, 6520 (1994). 50. J.M. Geary, J.W. Goodby, A.R. Kmetz, and J.S. Patel, J. Appl. Phys., 62(10), 4100 (1987). 51. E. Lee, P. Vetter, T. Miyashita, and T. Uchida, Jpn. J. Appl. Phys., 32, L1339 (1993). 52. S. Kuniyasu, H. Fukuro, S. Maeda, K. Nakara, M. Nitta, N. Ozaki, and S. Kobayashi, Jpn. J. Appl. Phys., 27(5), 827 (1988). 53. N. van Aerle, J. of the SID, 2(1), 41 (1994). 54. S. Ishihara, H. Wakemoto, K. Nakazima, Y. Matsuo, Liquid Crystals, 4(6), 669 (1989). 55. D. Seo, H. Matsuda, T. Oh-Ide, and S. Kobayashi, Mol. Cryst. Liq. Cryst., 224, 13 (1993). 56. N.P. Kuznetsov and M.I. Bessonov, Vysokomol. soyed., A28 (1), 100 (1986). 57. M. Ree, T.L. Nunes, and D.P. Kirby, Polym. Prep., 33(1), 309 (1992). 58. T.L. St. Clair and A.K. St. Clair, J. Polym Sci: Polym. Chem., 15, 1529 (1977). 59. P.M. Hergenrother, N.T. Wakelyn, and S.J. Havens, J. Polym Sci: Polym. Chem., 25, 1093 (1987). 60. P.M. Hergenrother and S.J. Havens, J. Polym Sci: Polym. Chem., 27, 1161 (1989). 61. C.R. Gautreaux, J.R. Pratt, and T.L. St. Clair, J. Polym. Sci.: Polym. Phys., 30, 71 (1992). 62. D.C. Rich, P. Huo, C. Liu, and P. Cebe, ACS Polym. Mat. Sci. Eng., 68, 124 (1993). 63. J. Friler and P. Cebe, Polynm. Eng. Sci., 33(10), 588 (1993). 64. P.P. Huo and P. Cebe, Polymer, 34(4), 696 (1993). 65. P.P. Huo, J. Friler, and P. Cebe, Polymer, 34(21), 4387 (1993). 66. J. T. Muellerleile, B. G. Risch, D. E. Rodrigues and G. Wilkes. Polymer, 34(4), 789 (1993). 67. R.A. Dine-Hart and W.W. Wright, Die Makromolek. Chem., 143, 189 (1971). 68. B.V. Kotov, T.A. Gordina, V.S. Voishchev, O.V. Kolinov, and A.N. Pravednikov, Vysokomol. Soyed., A19(3), 614 (1977). 69. J.M. Salley, T. Miwa, and C.W. Frank, in MaterialsScience of High Temperature Polymers for Microelectronics, Mater. Res. Soc. Proc., 227, p.1 17, D.T. Grubb, I. Mita, and D.Y. Yoon, Eds. (1991). 70. T.L. St. Clair, in Polyimides, p. 58, D. Wilson, H.D. Stenzenberger, and P.M. Hergenrother, Eds., Blackie, Glasgow (1990). 71. YE. P. Krasnov, A.YE. Stepanyan, Yu.I. Mitchenko, Yu.A. Tolkachev, and N.V. Lukasheva, Vysokomol. Soyed., A19 (7), 1566 (1977). 72. M. Fryd, in Polyimides: Synthesis, Characterization,and Applications, K.L. Mittal, Ed., Plenum Press, New York (1984). 73. A.K. St. Clair, T.L. St. Clair, W.S. Slemp, and K.S. Ezzell, "Optically Transparent/Colorless Polyimides", NASA Technical Memorandum 87650 (1985). 74. A.K. St. Clair, T.L. St. Clair, and W.S. Slemp, in Polyimides: Materials, Chemistry, and Characterization,p. 16, C. Feger, M.M. Khojasteh, and J.E. McGrath, Eds., Elsevier, Amsterdam (1989). 75. P.G. de Gennes, The Physics of Liquid Crystals, Ch.3, Oxford University Press, London (1975). 76. F.C. Frank, in Liquid Crystals: Their Physics, Chemistry, andApplications, p. 1, C. Hilsum and E. Raynes, Eds., Royal Society of London (1983). 77. J. Cognard, Mol. Cryst. Liq. Cryst., 78, Suppl. 1 (1982). 78. T. Uchida and H. Seki, in Liquid Crystals:Applications and Uses, B. Bahador, Ed., Vol. 3, Ch. 5., World Scientific, Teaneck, NJ (1990). 79. J. Castellano, Mol. Ciyst. Liq. Cryst., 94, 33 (1983). 80. S. Kobayashi, Y. limura, and M. Nishikawa, Conf. Rec. SID Int. Displ. Res. Conf, 78 (1994). 81. O'Mara, Liquid CrystalFlatPanelDisplays ManufacturingScience and Technology, Van Nostrand Reinhold, New York (1992). 82. R. Selvaraj, H. Lin, and J. McDonald, J. Lightwave Tech., 6(6), 1034 (1988). 83. J. Bandrup and E. Immergut, Eds., Polymer Handbook, John Wiley & Sons, New York (1989). 84. W. Groh and A. Zimmermann, Macromol., 24, 6660 (1991). 85. S. Saski, T. Matsuura, S. Nishi, and S. Ando, in Materials Science of High Temperature Polymers for Microelectronics, D.T. Grubb, I. Mita, and D.Y. Yoon, Eds., Mater. Res. Soc. Proc., 227 (1991). 86. J.W. Labadie, M.I. Sanchez, Y.Y. Cheng, and J.L. Hedrick, ibid., p. 43. 87. H. Haider, E. Chenevey, R.H. Vora, W. Cooper, M. Glick, and M. Jaffe, ibid., p. 35. 88. D.M. Stoakley and A.K. St. Clair, in Opticaland ElectricalPropertiesof Polymers, J.A. Emerson and J.M. Torkelson, Eds., Mater. Res. Soc. Proc., 214, 1990. 89. A.K. St. Clair, T.L. St. Clair, and W.P. Winfree, Polym. Mat. Sci. Eng., 59, 28 (1988). 90. G. Hougham, G. Tesoro, and J. Shaw, in Polyimides: Materials, Chemistry, and Characterization,p. 465, C. Feger, M.M. Khojasteh, and J.E. McGrath, Eds., Elsevier, Amsterdam (1989). 91. A. Misra, G. Tesoro, and G. Hougham, Polymer, 33(5), 1078 (1992). 92. T. Matsuura, Y Hasuada, S. Nishi, and N. Yamada, Macromol., 24, 5001 (1991). 93. T. Matsuura, M. Ishizawa, Y Hasuada, and S. Nishi, Macromol., 25, 3540 (1992). (1990). 94. D.M. Stoakley and A.K. St. Clair, in Polymeric Materialsfor Electronic Packaging and Interconnection, p. 87, J.H. Lupinski and R.S. Moore, Eds., ACS, Washington (1989). 95. D.L. Goff, E.L. Yuan, H. Long, and H.J. Neuhaus, ibid., p. 93. 96. A.K. St. Clair and T.L. St. Clair, in Polymers for High Technology - Electronics and Photonics, p.437, M.J. Bowden and S. R. Turner, Eds., ACS, Washington (1987). 97. T. Inchino, S. Sasaki, T. Matsuura, and S. Nishi, J. Polm. Sci: Polym Chem., 28, 323 (1990). 98. R. Reuter, H. Franke, and C. Feger, Applied Optics, 27(21), 4565 (1988). 99. G. Hougham, G. Tesoro, A. Viehbeck, and J. Chapple-Sokol, Polym. Prep, 34(1), 371 (1993). 100. D.W. Van Krevelen, Properties of Polymers: Their Correlation with Chemical Structure, Their Numerical Estimation and Prediction from Additive Group Contributions, Ch. 10, Elsevier, Amsterdam (1990). 101. J. Bicerano, Predictionof Polymer Properties, Ch. 8, Marcel Dekker, New York (1993). 102. M. Randic, J. Amer. Chem. Soc., 97(23), 6609 (1975). 103. L. Kier, L. Hall, W. Murray, and M. Randic, J. Pharm. Sci., 64(12), 1971 (1975). 104. J. Gladstone and T. Dale, Trans. Roy. Soc. (London), 148, 887 (1858). 105. A. Vogel, W. Cresswell, and I. Leicester, J. Phys. Chem., 58, 174 (1954). 106. R. Stein and G. Wilkes, in Structure and Propertiesof Oriented Polymers, I. Ward, Ed., J. Wiley, New York (1975). 107. G. Meeten, in Optical Properties of Polymers, Ch. 1, G. Meeten, Ed., Elsevier Applied Science, New York (1986). 108. P. Tien, R. Ulrich, and R. Martin, Appl. Phys. Lett., 14(9), 291 (1969). 109. G. Leclerc and A. Yelon, Appl. Optics, 23(16), 2760 (1984). 110. F. McCrackin, E. Passaglia, R. Stromberg, and H. Steinberg, J. NationalBureau Standards,67A(4), 363 (1963). 111. M. Born and E. Wolf, Principles of Optics Fifth Edition, p. 694, Pergamon Press, New York (1975). 112. C. Viney, Transmitted PolarisedLight Microscopy, Mc Crone Research Institute, Chicago (1990). 113. H. Jerrard, J. Opt. Soc. Am., 44(8), 634 (1954). 114. G. Baur, V. Wittwer, and D. Berreman, Phys. Lett., 56A(2), 142 (1976). 115. T.J. Scheffer and J. Nehring, J. Appl. Phys., 48(5), 1783 (1977). 116. M. Becker, R. Kilian, B. Kosmowski, and D. Mlynski, Mol. Cryst. Liq. Cryst., 132, 167 (1986). 117. B. Myrvold, K. Kondo, and S. Oh-Hara, Liquid Crystals, 15(3), 429 (1993). 118. B. Myrvold and K. Kondo, Liquid Crystals, 18(2), 271 (1995). 119. H. Fukuro and S. Kobayashi, Mol. Cryst. Liq. Cryst., 163, 157 (1988). 120. D. Berreman, Phys. Rev. Lett., 28(26), 1683 (1972). 121. D. Berreman, Mol. Cryst. Liq. Cryst., 23, 215 (1973). 122. U. Wolff, W. Greubel, and H. Kruger, Mol. Cryst. Liq. Cryst., 23, 187 (1973). 123. E. Matsui, K. Nito, and A. Yasuda, Liquid Crystals, 17(3), 311 (1994). 124. E. Lee, P. Vetter, T. Miyashita, T. Uchida, M. Kano, M. Abe, and K. Sugawara, Jpn. J. Appl. Phys., 32(10A), L1436 (1993). 125. Y. Kawata, K. Takatoh, M. Hasegawa, and M. Sakamoto, Liquid Crystals, 16(6), 1027 (1994). 126. Y. Zhu, Z. Lu, F. Quian, X. Yang, and Y. Wei, Appl. Phys. Lett., 63(25), 3433 (1993). 127. K. Kawahara, Y. Nakajima, T. Udagawa, H. Fujii, and H. Morimoto, Rec. SID Int. Disp. Res. Conf., 180 (1994). 128. Y. Nakajima, K. Saito, M. Murata, and M. Uekita, Mol. Cryst. Liq. Cryst., 237, 111 (1993). 129. M. Schadt, K. Schmitt, V. Kozinkov, and V. Chigrinov, Jpn. J. Appl. Phys., 31(7), 2155 (1992). 130. S. Jain and H. Kitzerow, Appl. Phys. Lett., 64(22), 2946 (1994). 131. S. Jain and H. Kitzerow, Jpn. J. Appl. Phys., 33(5A), L656 (1994). 132. L. Minsk, J. Smith, W. van Deusen, and J. Wright, J. Appl. Polym. Sci., 2(6), 302 (1959). 133. K. Ichimura, Y. Hayashi, K. Goto, and N. Ishizuki, Thin Solid Films, 235, 101 (1993). 134. K. Ichimura, Y. Hayashi, H. Akiyama, and T. Ikeda, Appl. Phys. Lett., 63(4), 449 (1993). 135. M. Sakuragi, T. Tamaki, T. Seki, Y. Suzuki, Y. Kawanishi, and K. Ichimura, Chem. Lett., 1763 (1992). 136. Z. Sekkat, M. Buchel, H. Menzel, and W. Knoll, Chem. Phys. Lett., 220, 497 (1994). 137. Z. Sekkat and M. Dumont, Synthetic Metals, 54, 373 (1993). 138. S. Sun, W. Gibbons, and P. Shannon, Liquid Crystals, 12(5), 869 (1992). 139. P. Shannon, W. Gibbons, and S. Sun, Nature, 368, 532 (1994). 140. W. Gibbons, P. Shannon, and S. Sun, and B. Swetlin, Nature, 351, 49 (1991). 141. W. Gibbons, A. Bell, P. Shannon, and S. Sun, Rec. SID Int. Disp. Res. Conf., 157 (1994). 142. W. Gibbons, P. Shannon, and S. Sun, SPIE, 1665, 184 (1992). 143. M. Hasegawa and Y. Taira, Rec. SID Int. Disp. Res. Conf., 213 (1994). 144. H. Kambe, in Aspects of Degradationand Stabilization of Polymers, Ch. 8, H. Jellinek, Ed., Elsevier, Amsterdam (1978). 145. R. Srinivasan and B. Braren, J. Polym. Sci.: Polym. Chem., 22, 2601 (1984). 146. G. Pettit and R. Suaerbrey, Appl. Phys. Lett., 58(8), 793 (1991). 147. R. Srinivasan, B. Braren, and R. Dreyfus, J. Appl. Phys., 61(1), 372 (1987). 148. J. Brannon, J. Lankard, A. Baise, F. Bums, and J. Kaufman, J. Appl. Phys., 58(5), 2036 (1985). 149. G. Pettit, M. Ediger, D. Hahn, B. Brinson, and R. Sauerbrey, Appl. Phys. A, 58, 573 (1994). 150. M. Schadt and W. Helfrich, Appl. Phys. Lett., 18, 127 (1971). 151. T.J. Scheffer and J. Nehring, Appl. Phys. Lett., 45(10), 1021 (1984). Chapter 2. Refractive Indices of Polyimides 2.1. Introduction The optical index of refraction is an important physical property of polyimides. The purpose of this chapter is to explore the fundamental factors that determine the indices of refraction of polyimides as a class of materials. The work focuses primarily on the relationship between polyimide chemistry and refractive index, but also investigates various processing effects. Refractive index is a measure of the material's electronic polarizability, which is an important component of its dielectric behavior. In many electronic packaging applications, a low dielectric constant is required for minimizing crosstalk and maximizing signal propagation speed. Refractive index is related to the dielectric constant by: Eo = n2 (1) where n is refractive index and eo is the dielectric constant at optical frequency. The index of refraction at optical frequencies, unlike the dielectric constant at electrical frequencies, is easily obtained experimentally, even if the material is anisotropic. The index of refraction is also important as a waveguide parameter. Polyimide electronic packaging layers have been suggested to serve as waveguides for optical interconnect technologies [1]. It is well known that polyimides with high fluorine content [2-19] have low dielectric constants and refractive indices. Unfortunately, however, there is no adequate explanation in the literature as to why this is so. There are three possible explanations: reduced crystallinity, reduced electronic polarizability, and reduced charge transfer complexation. First, it is possible that bulky fluorinated groups impede crystallization in polyimides. Index of refraction is related to density as shown in the Lorentz-Lorenz equation shown below: [(n2 - 1) / (n2 +2)] = (p/M) (Nav/3eo) x (2) where M is molecular weight, p is density, Nay is Avagadro's number, so is relative permittivty in a vacuum. In equation (2), for the range of refractive indices typical of organic polymers, refractive index increases with (p/M) and with a. Crystalline polymers are more densely packed and would therefore, in general, be expected to have higher indices of refraction. Though crystallinity has been observed in some polyimides [20-29], polyimides are most frequently considered to be primarily amorphous. Thus, reduced crystallinity is probably not a fully adequate explanation for the reduced indices of refraction in fluorinated polyimides. Another possible explanation is reduced electronic polarizability. Refractive index is related to electronic polarizability a as previously stated in equation (2). The carbonfluorine bond has low electronic polarizabilty, and consequently, fluorinated polymers are known to have low indices of refraction [30]. A third explanation is the reduced charge transfer complexation associated with the incorporation of bulky fluorinated groups. The groups sterically hinder the intermolecular interactions in the polyimide. This theory was put forth by St. Clair et al. [2,10-12]. It is quite possible that all three of these explanations play some role in the reduced indices of refraction. However, it has been entirely unclear as to what are the relative contributions of these effects. Discussions of this issue in the literature are frustratingly inadequate. For example, Inchino et al. [3] systematically increased the length of fluorinated alkoxy side chains in order to systematically reduce the dielectric constant of polyimides. The authors state, in agreement with St. Clair et al. [2,10-12], "the increase in length of the fluorinated alkoxy groups, leading to a more bulky separator which sterically hinders the [chain-chain] interaction, must reduce the dielectric constant " [3]. The authors then continue to state that "fluorine has low electronic polarizability which reduces the dielectric constant" [3]. However, it is unclear in this work what they believe is the main contributor. The explanation put forth by Matsuura et al. [15,16] for the reduced dielectric constant in rigid rod fluorinated polyimides is more confusing. According to the authors, "the low dielectric constants of the fluorinated polyimides result from reduced chain-chain interaction due to the low electronic polarizability of the fluorine" [16]. This is, of course, a non-sensical explanation, as the low electronic polarizability of the C-F bonds and the reduced intermolecular interactions are different and unnecessarily related phenomena. Hence, it is the goal of this chapter to gain better insight into the mechanisms of the refractive index reduction in polyimides that occurs as a result of fluorination. The optical properties of a large number of polyimides are compared. Experimental refractive indices are then compared to indices calculated by the property prediction techniques described in Chapter 1. The effects of processing on the index of refraction in polyimides are also investigated later in the chapter. 2.2. Experimental Techniques 2.2.1. Polyimide film preparation Two sources of polyimide refractive index data were used. The first was the literature [2-5,31-36]. Films preparation for these samples are described by the authors. The second source was polyimide films synthesized at and provided by Dr. Anne St. Clair of the NASA Langley Research Center, Hampton, Virginia. The chemical compositions of these and other polymers are presented in Tables 2.1-2.3. Polyimides ODPA-3,3'ODA, ODPA-4-BDAF, 6FDA-3,3'-ODA, 6FDA-4-BDAF, 6FDA-DASP, BFDA-DASP, BPDA-DASP, ODPA-DASP, IPAN-DASP, BTDA-DASP, HQDEA-DASP, BDSDADASP, BTDA-1,3BABB, PMDA-ODA, and ODPA-PDA were obtained as thick free standing films. The polyamic acids were synthesized in N,N-dimethylacetamide, solution cast on glass, and cured in a low humidity environment at 100 0 C, 2000 C, and 3000 C for one hour at each temperature. PMDA-ODA was annealed an additional 20 minutes at 4000 C. RegulusTM NEW-TPI, whose chemical composition is shown in Table 2.1, was received from the Mitsui Toatsu Chemical Co. Polyimides discussed in section 2.4.3, PMDA-ODA, PMDA-APB, PMDA-BDAF, 6FDA-ODA, 6FDA-APB, and 6FDA-BDAF, were received as 15% solids by weight solutions in N,N-dimethylacetamide. The solutions were spin coated on glass using a Headway Research spin coater at 4000 RPM for 30 seconds. The coated films were progressively cured at 100 0 C, 200 0 C, and 3000 C for one hour at each temperature. 2.2.2. Zone drawing The zone drawing technique has been described previously [37-40] and is depicted in Figure 2.1 below. The film is placed between two clamps with a weight pulling on one end. A zone heater passes over the film at a constant rate which heats the film above the glass transition temperature. The material is drawn at the site of heating. By varying the drawing weight and temperature, films of various draw ratios (final length / initial length) were produced for 6FDA-4-BDAF fluorinated polyimide films. Zone Heater >Tg I Grinp V Grips Polymer film Dead Weight Load Figure 2.1. Film zone drawing method. 2.2.3. Refractive index measurement Refractive index measurements were made for polyimide films using a Metricon PC-2000 prism coupler with X = 6328A. Average refractive indices were calculated as: navg = (1/3) (2nTE + nTM) (3) The prism coupling technique for measuring index of refraction is discussed in Chapter 1. 2.2.4. Refractive index prediction Two property prediction techniques, which are discussed in Chapter 1, were utilized in order to estimate how varying chemical make-up should affect refractive index. It is assumed that these estimation techniques account for the intrachain electronic polarizability of the polymer molecular repeat unit. The first is the group contribution method by van Krevelen [41] based on an equation used by Vogel [42]. The second is a topological technique by Bicerano [43]. Both methods, unlike two other techniques described by van Krevelen [41], require no measurement of the polymer density prior to the calculation. Only the chemical structure is needed. However, since polymer density is not used in these calculations, both methods are intended only for amorphous polymers. For details concerning these calculations, please refer back to Chapter 1 or the original references [41-43]. 2.3. Relationship Between Polyimide Chemistry and Refractive Index Experimental and predicted refractive indices of a large number of polyimides were compared. Three types of behavior were found. First, many amorphous, non-fluorinated polyimides were found to have refractive indices of around 1.68-1.70, deviating slightly, though systematically, above the predicted refractive index. Second, some polyimides were found to have refractive indices that were substantially higher than those predicted. Third, fluorinated polyimides had low refractive indices which generally compared favorably to predicted values. 2.3.1. Amorphous, non-fluorinated polyimides A comparison between predicted and measured refractive indices in nonfluorinated polyimides is shown in Figure 2.2. The polyimides have generally been arranged from left to right in order of declining measured refractive index. The chemical structures of the polyimides are shown in Table 2.1. Although the calculated refractive indices are close to experimental values, we notice a systematic underprediction, especially when using the presumably more accurate method of Bicerano. Refractive indices calculated by van Krevelen's method are good, underpredicting by only about 0.02-0.03. Bicerano's method underpredicts refractive index by about 0.05-0.06, which is three to four times the standard deviation in Bicerano's correlation. It is unlikely that such large deviations from experimental values in the Bicerano computations can be attributed solely to statistical error. Charge transfer complexation [2,10-12,44-49] and/or molecular ordering might result in deviations of predicted from measured refractive indices. There has been no evidence of substantial crystal lattice ordering in these materials. However, as shown in Figure 2.3(a), PMDA-ODA may exhibit slight paracrystalline order when examined using wide-angle X-ray scattering (WAXS). Alternatively, chain packing which might result from intermolecular complexation could be responsible for an increase in density of the amorphous phase and/or an increase in electronic polarizability. The property prediction techniques apparently do not account for the chain packing and/or charge transfer effects. 2.3.2. Semicrystalline polyimides The index of refraction of a polymer is directly related to its density. Since crystalline phases are denser than amorphous phases, a measured refractive index that is substantially higher than that predicted by property prediction methods may be associated with polymer semicrystallinity. Figure 2.4 compares predicted and measured indices of polyimides with unusually high refractive indices. See Table 2.2 for the chemical structures. There is substantial experimental evidence of crystal lattice ordering in these particular materials. Previous authors have shown that BPDA-PDA [22], ODPA-PDA [20], BTDA-1,3-BABB [21,23-27] are crystallizable. WAXS analysis, shown in Figure 2.3(b), confirms the presence of a crystalline phase in our ODPA-PDA sample. This material has exactly the same atoms and bonds as does PMDA-ODA (chemical structure Table 2.1, WAXS Figure 2.3(a)), but has a refractive index significantly greater. Thus, there is likely a strong structure-property relationship between crystallinity and refractive index in polyimides which is not taken into account in the van Krevelen and Bicerano predictions. 2.3.3. Polyimides containing hexafluoroisopropylidene (6F) The optical properties of polyimides containing the hexafluoroisopropylidene (6F) group have been the subject of much investigation [2-19]. It has generally been observed that the refractive indices of polyimides decrease with increasing fluorine content. It is also believed by these authors that the 6F group, which consists of two trifluoromethyl groups bonded to carbon, sterically inhibits charge transfer complexation in the polyimides. Figure 2.5 shows a comparison of predicted and measured refractive indices of polyimides containing the 6F group. For the chemical structures, see Table 2.3. Both property prediction methods estimate refractive index quite accurately. Bicerano's method is accurate when the 6F group appears in both the dianhydride and diamine portions of the repeat unit, but very slightly underpredicts refractive index when the 6F group exists in only one segment. This demonstrates that the lower refractive indices that have been observed experimentally in 6F materials can be predicted from the modified chemical structure. Furthermore, the good agreement between measured refractive index and these predictions suggests that chain packing and charge transfer complexation may be reduced in these materials. 2.3.4. Polyimides containing fluorinated alkoxy side chains The synthesis and characterization of novel polyimides containing fluorinated alkoxy side chains, shown in Table 2.4, have been reported [3]. Measured refractive indices and dielectric constants were shown to be reduced systematically with increasing fluorine content. The fluorine content of the polyimides was increased by increasing the length of the side chains. In Figure 2.6, predicted refractive index (Bicerano technique) is shown as a function of fluorine content. The predicted indices compare fairly favorably to measurement. However, experimentally, a slight dependence on the dianhydride chemistry was found by the authors. Polyimides containing the 6FDA dianhydride, for example, had lower refractive indices than polyimides containing non-fluorinated dianhydrides, but equal overall fluorine content. This is not predicted, however, by Bicerano's method, as shown in Figure 2.6. This may indicate that the 6F group is more effective per fluorine atom than is the side chain in reducing polymer chain packing. 2.3.5. Polyimides containing pentafluorosulfanyl (SF 5) Recently, St. Clair and St. Clair [19] reported the synthesis of novel fluorinated polyimides containing a pentafluorosulfanyl (SF5) group. Measured and predicted refractive indices of these polymers are shown in Table 2.5. We found a small deviation between predicted and experimental values. The least deviation occurred in a polyimide containing a bulky isopropylidene group in the dianhydride portion of the chain (IPAN). Likely, the SF5 group impedes chain packing to some extent, though not as well as the 6F group. 1.8I 9 £ Measured van Krevelen Ricerann •IAVrl "M·v 1.7: A a 0 0 O 0 D£ r O O 0 n v 0v 1.6 1.5 I II I I I -- I --- III I I 5 Polyimide Figure 2.2. Experimental and predicted refractive indices of non-fluorinated polyimides in Table 2.1. C C (I 10 20 30 40 TWO THETA (DEGREES) Figure 2.3. Wide angle X-ray scattering (WAXS) of: a.) PMDA-ODA and b.) ODPAPDA polyimide films. The polymers contain the same exact atoms and bonds, but have very different physical structures, and correspondingly, very different optical properties. Table 2.1. Dianhydride (R) and diamine (Z) elements of polyimides in Figure 2.2. 0 C -N O \ R 6 N-Z- 6 1.8-1 1.7 S0o O o 1.6 - A Measured 1 van Krevelen o Bicerano 1.5 Polyimide Figure 2.4. Measured and predicted refractive indices of semicrystalline polyimides in Table 2.2. 83 Table 2.2. Dianhydride (R) and diamine (Z) elements of polyimides in Figure 2.4. 84 A Measured O Van Krevelen o Bicerano 1.7 0 0A0 1.6 1 0 0 o~oos , 0 0 ooo go A at 0 0 i I I a A 1.5 I I I I I I I Polyimide Figure 2.5. Measured and predicted refractive indices of 6F polyimides from Table 2.3. E 0 71 (NJ 0 (N- Uo z> 0C H:E o *1 86 -- 1.8t x "O 1.7- - 0 I.. ,- O 6FDA C)C A BPDA 1.6- O BTDA 0o 1.5 "4 10. 1.4 SI 0 10 I 20 30 I I 40 50 60 Fluorine content (wt. %) Figure 2.6. Predicted refractive indices (Bicerano's method) of polyimides with fluorinated alkoxy side chains vs. fluorine content. Chemical structures in Table 2.4. Table 2.4. Dianhydride (R) and diamine (Z) segments of polyimides with fluorinated alkoxy side chains, reported by Inchino et al. [3]. R Z -- CF3 CH2 (CF2)3 F NJ 6FDA OCH 2 (CF2)7 F BPDA O II CH2 (CF2)6 F 4-'--C BTDA CH2 (CF2)10 H Table 2.5. Measured and predicted refractive indices of SF5 polyirnides. SFs O -N R C N. C0 Polyimide C OI 0 R CF63 6FDA-DASP CCF3 CF3 CF3 BFDA nmeas. 1.541 nBic. 1.536 nvK 1.529 1.576 1.559 1.554 1.640/ 1.592 1.582 DASP , BPDABPDADASP 1.635* ODPADASP 1.620 1.588 1.576 1.597 1.583 1.570 1.621 1.588 1.577 1.624 1.594 1.586 ,CH P 3 CHI3, IPAN-DASP O II BTDADASP C HQDEA- BDSDA- 0 Z O. Z.O . 1.647 1.616 1.602 * First value represents side 1, the second value, side 2. 2.4. Relationship Between Polyimide Processing and Refractive Index 2.4.1. In-plane birefringence in polyimide films PMDA-ODA PMDA-ODA is a polyimide that is widely used commercially. The repeat unit of the polymer is shown below: O O -N N C II C II O O Figure 1.3 shows three types of orientation in polymer films: isotropic orientation, in-plane orientation, and uniaxial orientation. Thin films of PMDA-ODA have been found to have" out-of-plane refractive indices, nTM, that are significantly lower than in-plane refractive indices, nTE [31,32]. This is because the main chain (c) axes lie preferentially in the plane of the film as shown in Figure 1.3(b). The average refractive index, navg = (2nTE + nTM) / 3, and birefringence, An = nTE- nTM, of fully annealed spun cast PMDA-ODA films were reported by Russell et al.[31]. The authors found navg = 1.695 and An = 0.08. Data for our PMDA-ODA samples are shown in Table 2.6. Solvent cast film received from the NASA Langley Research Center had a birefringence of 0.078, while commercial DuPont T Kapton m H had a birefringence of 0.111. The commercial film, as compared to the NASA Langley film, had a lower out-of-plane refractive index nrM and a higher in-plane refractive index nTE, which suggests that the film has a greater level of in-plane ordering. The average refractive indices, however, of both films were around 1.7. Table 2.6. Optical Properties of PMDA-ODA. Material source: nav NASA Langley 1.697 DuPont Kapton M H 1.702 inTE nTM An 1.723 1.645 0.078 1.739 1.628 0.111 Optically transparent polvimides Refractive indices of four optically transparent polyimides synthesized at the NASA Langley Research Center are listed in Table 2.7. Chemical structures can be found in Tables 2.1 and 2.3. Unlike the PMDA-ODA films, very little optical anisotropy is seen in these materials. We conclude that there is little preferential in-plane alignment of the polymer chains in these films, as shown in Figure 1.3(a). This may be a result of either the high thickness of the films or a lack of a tendency of these materials to order in the plane of the film. Table 2.7. Refractive indices of optically transparent polyimides. Name ODPA-3,3'-ODA ODPA-4-BDAF 6FDA-3,3'-ODA 6FDA-4-BDAF nTM 1.683 1.615 1.598 1.566 nTE 1.681 1.612 1.597 1.564 Thickness 23tm 28pm 34pm 34gm 2.4.2. Birefringence in zone drawn films Nakagawa [32] and Smirnova and Bessonov [34] report the optical properties of uniaxially cold drawn PMDA-ODA. They show that the refractive index is highest in the direction of drawing and lowest normal to the direction of orientation. This is because the polarizable groups in the main chain are preferentially oriented along the draw axis, as shown in Figure 1.3(c). Both authors report a maximum birefringence on the order of 0.20. The average refractive index, however, is unaffected by the level of orientation, as it is near 1.7, which is consistent with what was previously mentioned in Section 2.4.1. The chemical structure of 6FDA+4-BDAF, a fluorinated, low refractive index polyimide, is shown below: CF,3 ,CF3 When the material is zone drawn along one axis of the film, significant optical anisotropy is observed. Figure 2.7 shows the refractive index of 6FDA+4-BDAF in three perpendicular directions as a function of draw ratio. In the direction of the draw in the plane of the film, refractive index increases with draw ratio. Refractive index decreases with draw ratio, however, out of the plane of the film or in the plane of the film perpendicular to the direction of the draw. This is, again, a result of the preferential alignment of the primary polarizable groups along the main chain in the direction of the draw. The average refractive index is not affected by zone drawing. Refractive index is lowest out of the plane of the film. At high draw ratios, it is nearly as low as 1.54, which is about 0.017 less than the average index of the film. 1.60 It - A€" S1.58 a A in-plane/parallel to draw A aI. 1.56- E in-plane/normal to draw O 0 0 D o oO 0 out-of-plane 1.54 .1- 1 2 Draw ratio 3 Figure 2.7. Refractive index vs. draw ratio for 6FDA+4-BDAF in the direction of the draw and normal to the draw. 2.4.3. Effect of cure temperature on refractive index As explained in Chapter 1,polyimides generally begin as polyamic acid precursors, which are thermally converted to polyimides. As the polyimides undergo this reaction, the refractive index of the material increases. This is demonstrated in Figure 2.8, which shows the average indices of refraction of several polyimides as a function of cure temperature. The films were spin coated from solutions obtained from NASA Langley, and their chemical structures can be found in Tables 2.1-3. Note that, as discussed previously, refractive index decreases with increased fluorine content. To investigate the reasons for the increase in refractive index with increased cure temperature, the predicted refractive indices of the polyamic acid precursors were calculated using the Bicerano technique. In Figure 2.9(a), these values are compared to the measured refractive indices of the 100 0 C cured materials. Apparently, the values are in reasonably good agreement, regardless of whether the polyimide contains the 6F group. According to Kan and Kao [50], the imidization reaction does not occur appreciably until about 150 0 C, but occurs rapidly above 250 0C. It is thus reasonable to assume that the 100 0 C treatment is representative of the unconverted state, while the 3000C is representative of the imidized state. In Figure 2.9(b), predicted refractive indices of the fully converted polyimide are compared to measured refractive indices after the 300 0 C cure. As found in the previous sections, the Bicerano underpredicts the refractive indices of the non-fluorinated polyimides, i.e., PMDA-ODA and PMDA-APB, but is in good agreement with the refractive indices of the 6F fluorinated polyimides, i.e., the remaining four polymers. The good agreement with predicted values for the polyamic acids suggests that there are no unexpected chain packing or charge transfer effects in the precursor materials that go unaccounted for by the Bicerano prediction technique. However, when the materials are converted to polyimide, the non-fluorinated materials undergo a significant unpredicted increase in refractive index. Since this increase in refractive index cannot be accounted for based upon the chemical change alone, this study suggests that the imidization reaction leads to greater chain packing and/or increased interchain interaction in the non-fluorinated polyimides. The 6F materials, on the other hand, remain consistent with predicted values upon imidization, indicating that the 6F group impedes the chain packing. 1.75 I ~ ~I~ IC_ _ 1_1_ 1.70- PMDA-ODA --e--- 1.65- 1.60~--=s~-~ ~==;--~ft===z~~ 1 _ ~--~-~t---------6 -I - 1.55 1.50 rl 1.50- I ---100 200 Temperature ----- PMDA-BDAF -- 6FDA-ODA 6--- ---- o-- 6FDA-APB -- 6FDA-BDAF *-- -- - --- PMDA-APB - - 300 (0C) Figure 2.8. Refractive indices of several polyimides as a function of cure temperature. -1 a A 100 0 C (exp.) o Bicerano (calc.) 1.7 1.60o b0 . .. . i. . .. . . Polyamic Acid Figure 2.9. Predicted (Bicerano's method) vs. measured refractive indices of several polyimides. a.) Predicted values of the polyamic acid precursors vs. measured values of the films after only 100 0Ccure. 3 b 3000 C (exp.) Bicerano (calc.) A o 1.7 o 1.6 o A 0 0 2 Polyimide Figure 2.9. continued. b.) Predicted values of the polyimides vs. measured values after the 300 0 C cure. 2.4.4. Effect of post-cure annealing on NEW-TPI polyimide Regulus TM NEW TPI [28,29] is a polyimide that is known to crystallize when annealed above its glass transition temperature (Tg = 2500 C). Crystallinity, as previously discussed, can increase the index of refraction of a polymer. Several samples of fully cured NEW-TPI were annealed in order to crystallize the material. Films were treated for 10 minutes at various temperatures between 2800 C and 3200 C, and for various times at 2900 C. The annealing process roughened the surface of the films, thus making it more difficult to take accurate data using the Metricon prism coupler. Each refractive index (nTE or nTM) is therefore the average of four measurements. Both the average refractive index and birefringence (nTE - nTM) increased with annealing temperature, as shown in Figure 2.10. There is a significant step in average refractive index and birefringence at 2900 C, indicating that this could be the onset temperature of rapid crystallization. The increase in bulk birefringence suggests that the crystals may have preferentially order in the plane of the film. The refractive index and birefringence also increase with annealing time at 2900 C, as shown in Figure 2.11. Average refractive index appears to begin increasing after about three minutes of annealing, appears to be about half way to its maximum near six minutes, and appears to level after about 10 minutes. Previously reported work [28] for NEW-TPI samples isothermally annealed at 300 0 C, indicates a crystallization half time (ti/2) of about 6.3 minutes and a crystallization completion time (tc) of about 12 minutes. The general increase in birefringence with annealing time also suggests that the crystals may have inplane order. 2.5. Conclusions Significant variations of the refractive index of polyimides have been observed with varying, often subtly varying, chemical structure. Property prediction calculations of refractive indices of polyimides give some insight into the causes of these changes. While charge transfer complexation effects may slightly increase the refractive indices of polyimides, semicrystallinity or short range ordered structure can increase the refractive index substantially. Incorporation of fluorinated groups is an effective means of reducing the refractive index. The observed reduction results from a combination of reduced electronic polarizability and reduced interchain interaction and/or chain packing. Refractive (a) -1. *o W 1 •= 1. .z3 I 1W tW I. -·I As rec 300 290 280 310 320 Temperature (OC) (b) 0.025 0.020 - 0.015 0.010 0.005·.·. :i: :::::::::;::::: 0.000 :.:: :::: I6-"""w - As rec 280 290 = Mý """""*** 300 * "0 ...WOMMIN. bw """""" 310 320 Temperature (OC) Figure 2.10. a.) Average refractive index of NEW-TPI vs. post-cure annealing temperature. b.) Birefringence vs. post-cure annealing temperature. 1.72 1.71 1.70 1.69 1.68 1 .( I I I I I 0 1 2 3 4 I 5 6 Minutes at I I I I 7 8 9 0 290 C 10 11 12 8 0 290 C 10 (b) 0.025 0.020 0.015 0.010 0.005 0.000 0 2 4 6 Minutes at 12 Figure 2.11. a.) Average refractive index vs. post-cure annealing time at 290 0C. b.) Birefringence vs. post-cure annealing time index prediction techniques appear to be highly accurate for polyimides with bulky fluorinated substituents, but underpredict for non-fluorinated polyimides. In addition to chemical structure, polyimide processing can affect the index of refraction. Oriented materials, for example, will be birefringent. While uniaxially drawing the films will, of course, induce orientation, birefringence can sometimes be induced simply by the casting or annealing process. The overall refractive index can also be dependent upon the annealing conditions, as both amic acid to imide conversion and polymer crystallization increase the refractive index. 2.6. References 1. R. Selvaraj, H.T. Lin, and J.F. McDonald, J. Lightwave Tech., 6(6),1034 (1988). 2. A.K. St. Clair and T.L. St. Clair, in Polymersfor High Technology - Electronicsand Photonics, p. 437, M.J. Bowden and S. R. Turner, Eds., ACS, Washington (1987) 3. T. Inchino, S. Sasaki, T. Matsuura, and S. Nishi, J. Polm. Sci: Polym Chem., 28, 323 (1990). 4. R. Reuter, H. Franke, and C. Feger, Applied Optics, 27(21), 4565 (1988). 5. G. Hougham; G. Tesoro; A. Viehbeck; J. Chapple-Sokol, Polym. Prep, 34(1), 371 (1993). 6. S. Saski, T. Matsuura, S. Nishi, and S. Ando, in Materials Science of High Temperature Polymers for Microelectronics,D.T. Grubb, I. Mita, and D.Y. Yoon, Eds., Mater. Res. Soc. Proc., 227 (1991). 7. J.W. Labadie, M.I. Sanchez, Y.Y. Cheng, and J.L. Hedrick, ibid., p. 43. 8. H. Haider, E. Chenevey, R.H. Vora, W. Cooper, M. Glick, and M. Jaffe, ibid., p. 35. 100 9. D.M. Stoakley and A.K. St. Clair, in Optical and ElectricalPropertiesof Polymers, J.A. Emerson and J.M. Torkelson, Eds., Mater. Res. Soc. Proc., 214 (1990). 10. A.K. St. Clair, T.L. St. Clair, and W.P. Winfree, Polym. Mat. Sci. Eng., 59, 28 (1988). 11. A.K. St. Clair, T.L. St. Clair, W.S. Slemp, and K.S. Ezzell, "Optically Transparent/Colorless Polyimides", NASA Technical Memorandum 87650 (1985). 12. A.K. St. Clair, T.L. St. Clair, and W.S. Slemp, in Polyimides: Materials, Chemistry, and Characterization,p. 16, C. Feger, M.M. Khojasteh, and J.E. McGrath, Eds., Elsevier, Amsterdam (1989). 13. G. Hougham, G. Tesoro, and J. Shaw, ibid. p. 465. 14. A. Misra, G. Tesoro, and G. Hougham, Polymer, 33(5), 1078 (1992). 15. T. Matsuura, Y Hasuada, S. Nishi, and N. Yamada, Macromol., 24, 5001 (1991). 16. T. Matsuura, M. Ishizawa, Y Hasuada, and S. Nishi, Macromol., 25, 3540 (1992). (1990). 17. D.M. Stoakley and A.K. St. Clair, in Polymeric Materialsfor Electronic Packaging and Interconnection, p.87, J.H. Lupinski and R.S. Moore, Eds., ACS, Washington, (1989). 18. D.L. Goff, E.L. Yuan, H. Long, and H.J. Neuhaus, ibid., p. 93. 19. A.K. St. Clair and T.L. St.Clair, Polym. Prep., 34(1), 385 (1993). 20. M.K. Gerber, J.R. Pratt, and T.L. St. Clair, in MaterialsScience of High TemperaturePolymersfor Microelectronics,p. 487, D.T. Grubb, I Mita, and D.Y. Yoon, Eds., Mater. Res. Soc. Proc., 227 (1991). 101 21. P.M. Hergenrother and S.J. Havens, in MaterialsScience of High Temperature Polymersfor Microelectronics,p. 453, D.T. Grubb, I Mita, and D.Y. Yoon, Eds., Mater. Res. Soc. Proc., 227 (1991). 22. M. Ree, T.L. Nunes, and D.P. Kirby, Polym. Prep., 33(1), 309 (1992). 23. P.M. Hergenrother and S.J. Havens, J. Polym. Sci. A: Polym. Chem., 27, 1161 (1989). 24. P.M. Hergenrother, N.T. Wakelyn, and S.J. Havens, J. Polym. Sci. A: Polym. Chem., 25, 1093 (1987). 25. D.C. Rich, P. Huo, C. Liu, and P. Cebe, ACS Polym. Mat. Sci. Eng., 68, 124 (1993). 26. J. Friler and P. Cebe, Polym. Eng. Sci., 33(10), 588 (1993). 27. J. T. Muellerleile, B. G. Risch, D. E. Rodrigues and G. Wilkes. Polymer, 34(4), 789 (1993). 28. P.P. Huo and P. Cebe, Polymer, 34(4), 696 (1993). 29. P.P. Huo, J. Friler, and P. Cebe, Polymer, 34(21), 4387 (1993). 30. W. Groh and A. Zimmermann, Macromol., 24, 6660 (1991). 31. T.P. Russell, H. Gugger, and J.D. Swallen, J. Polym. Sci: Polym. Phys., 21, 1745 (1983). 32. K. Nakagawa, J. Appl. Polym. Sci., 41, 2049 (1990). 33. S. Noe, Ph.D. Thesis, Massachusetts Institute of Technology (1992). 102 34. V.E. Smirnova and M.I. Bessonov, in Polyimides: Materials, Chemistry, and Characterization,edited by C. Feger, M.M. Khojasteh, and J.E. McGrath, Elsevier, Amsterdam, 1989, p. 563. 35. S. Herminghaus, D. Boese, D.Y. Yoon, and B.A. Smith, Appl. Phys. Lett., 59(9), 1043 (1991). 36. D. Boese, H. Lee, D.Y. Yoon, J.D. Swallen, and J.F. Rabolt, J. Polym. Sci.: Polym. Phys., 30, 1321 (1992). 37. T. Kinugi, C. Inchinose, and A. Suzuki, J. Appl. Polym. Sci., 31, 429 (1986). 38. Y. Aihara and P. Cebe, Polym. Prepr., 33(1), 471 (1992). 39. Y. Aihara and P. Cebe, Polym. Prepr., 33(2), 633 (1992). 40. J.B. Teverovsky, D.C. Rich, Y. Aihara, and P. Cebe, J. Appl. Polym. Sci., 54, 497 (1994). 41. D.W. van Krevelen, Propertiesof Polymers, Elsevier, Amsterdam (1990). 42. A. Vogel, W. Cresswell, and I. Leicester, J. Phys. Chem., 58, 174 (1954). 43. J. Bicerano, Predictionof Polymer Properties,Marcel Dekker, New York (1993). 44. R.A. Dine-Hart and W.W. Wright, Die Makromolek. Chem., 143, 189 (1971). 45. B.V. Kotov, T.A. Gordina, V.S. Voishchev, O.V. Kolinov, and A.N. Pravednikov, Vysokomol. Soyed., A19(3), 614 (1977) 46. J.M. Salley, T. Miwa, and C.W. Frank, in Materials Science of High Temperature Polymers for Microelectronics,p. 117, D.T. Grubb, I Mita, and D.Y. Yoon, Eds., Mater. Res. Soc. Proc., 227 (1991). 103 47. T.L. St. Clair, in Polyimides, D. Wilson, H.D. Stenzenberger, and P.M. Hergenrother, Eds., Blackie, Glasgow (1990). 48. YE. P. Krasnov, A.YE. Stepanyan, Yu.I. Mitchenko, Yu.A. Tolkachev, and N.V. Lukasheva, Vysokonmol. Soyed., A19 (7), 1566 (1977). 49. M. Fryd, in Polyimnides: Synthesis, Characterization,and Applications, K.L. Mittal, Ed., Plenum Press, New York (1984). 50. L. Kan and K. Kao, J. Chem. Phys., 98(4), 3445 (1993). 104 Chapter 3. Curing Study of Probimide® 412 Preimidized Photosensitive Polyimide 3.1. Introduction Over the past decade and a half, there has been substantial interest in photosensitive polyimides (PSPIs) [1]. The photosensitive behavior allows the polyimides, which are widely used in electronics applications, to be patterned photolithographically. The first type of PSPI developed was a photosensitive polyamic ester [2], which crosslinks only in the precursor state. The second type was an intrinsically photosensitive polyimide [3,4] which is both solvent soluble and photosensitive in its fully converted state. It is the latter version of the PSPI with which we are concerned. In general, polyimides require high temperature cures for conversion of the precursor polymer and for the removal of solvent. The chemical and physical properties of polyimide systems change substantially as they are cured, and the material is converted from precursor to polyimide. Because the intrinsically photosensitive polyimides are already in their fully converted state, and because they gel upon exposure to ultraviolet light, they show strong potential as high performance materials which can be cured rapidly and at low temperatures. One application in which rapid, low temperature curing is highly desirable is in their use as alignment films for liquid crystal display devices [5,6]. There has been much work published regarding the performance and physical properties of intrinsically PSPI films that have undergone ultraviolet exposure and conventional high temperature thermal cures [7-14]. Ree et al. [14] found that chemical crosslinks are formed by both photochemical and thermal curing. In the present work, we show how thermal and ultraviolet curing, which do not involve closure of the imide ring, physically and chemically change the material. We do so by examining a series of films that have undergone step cures from soft bake to hard bake. We report the effects of the atmosphere in which the hard bakes are performed, and the effects of ultraviolet exposure on the films. 105 3.2. Experimental 3.2.1. Polymer film samples The polyimide system studied is of the Probimide® 400 series produced by OCG Microelectronic Materials, Inc. The polymer contains the following photosensitive structural unit: 0 _l Isb u P -N I 0II where R, R' = alkyl group. The polyimide contains a benzophenone (Ph-CO-Ph) group in the dianhydride portion of the repeat unit. The polyimide is obtained in y-butyrolactone (GBL) solvent, and the product used in this study, Probimide® 412, has a 12.5% original solids content by weight. Bulk films of Probimide® 412 on the order of 100 gpm thick were produced. All films were amorphous, and exhibited no Bragg scattering peaks when examined by wide angle X-ray scattering. The film casting procedure is depicted in Figure 3.1. The polyimide solution was first dropped onto a glass plate and wiped with a metal doctor blade with a 1 mm gap as shown in Figure 3.1(a). The wet films were dried to a tack free state under a dry air current at room temperature overnight as shown in Figure 3.1(b). The parent films were then systematically treated at higher temperatures. Samples were cut from the film after each curing step. The first treatment was 4.5 hours at 100 0C in air. The second was 3 hours at 200 0 C in air. Films were then treated at 300 0 C for three hours in either air or nitrogen. Material treated at 3000C in nitrogen was then additionally cured at 4000 C for one hour in nitrogen. Any Probimide® 412 material treated at 400 0C in air was essentially destroyed by oxidation. A separate film was treated at 140 0 C for 3 hours in air after the 4.5 hour bake at 100 0C. These thermally cured samples were not exposed to ultraviolet irradiation. Material that had undergone the 4.5 hour treatment at 100 0C was irradiated under a Spectroline model BIB-150B ultraviolet lamp with X = 364 nm and an area normalized power of approximately 2.5 mW/cm 2. The film was exposed to approximately 27 J/cm 2 106 over the course of 3 hours, each side of the film being exposed for 90 minutes. In order to separate the effects of hard bake and UV cures, we did not subsequently hard bake the irradiated film. Thin films of Probimide® 412 were spun cast from the original solution in GBL on glass or KBr at 4000 RPM for 30 seconds. The samples were immediately softbaked at 100 0C for fifteen minutes in air. The samples appeared dry after softcuring. The thin films were then progressively cured at 2000 C, 300 0C, and 400 0C, one hour at each temperature. The 200 0C cures were performed in air, while the 3000 C cures were performed in either air or nitrogen. The 400 0 C cures were performed in nitrogen only on films previously treated at 3000 C in nitrogen. Spin cast films treated at 100 0C were irradiated under the UV lamp for various times. 3.2.2. Measurement techniques A Seiko TG/DTA320 was used for thermogravimetric analysis (TGA) of the bulk films. Samples were scanned from room temperature to 575 0 C at 100C/minute in flowing air (300 ml/minute). Dynamic mechanical analysis (DMA) was performed in tension mode on bulk films using a Seiko DMS200. Samples were scanned under nitrogen at 20C/minute from room temperature to the temperature above softening at which breakage occurred. Frequencies 1, 2, 5, 10, and 20 Hz were used. Only the 1 Hz data is presented here, as the frequency dependence of the DMA spectra was unnoteworthy. Index of refraction at X = 632.8 nm was measured using a Metricon PC-2000 prism coupler. The prism coupler technique is discussed in Chapter 1. Refractive index was measured by determination of the waveguide propagation modes. Both in-plane (nTE) and out-of-plane (nTM) indices were measured. The average refractive index navg was calculated as shown below: navg = (1/3) (2nTE + nTM) (1) Refractive index measurements were taken for the free standing films (by critical angle) and the films spin coated on glass (by waveguide mode determination). Fourier transform infrared spectroscopy (FTIR) was performed using a Nicolet 510P. FTIR spectra were taken for the Probimide® 412 films spin coated on KBr. 107 SGlass plate blad Dotor with 1mm gap (b) Glass plate with coated polymer film Figure 3.1. Prepartion of bulk films. a.) Coating and blading process for optimization of film smoothness and thickness uniformity. b.) Pre-cure air-drying chamber. 108 3.3. Results And Discussion 3.3.1. Coloration And Brittleness The changes in coloration and brittleness in Probimide® 412 thick films that occur with subsequent cures are summarized in Table 3.1. Probimide® 412 is slightly yellow and flexible when cured at 1000 C. These characteristics change little upon curing to 2000 C. The material becomes dark orange and brittle when cured at 300 0 C in air, but remains yellow and flexible in nitrogen. However, the polymer turns brown and brittle when treated at 400 0C in nitrogen. The dark coloration and brittleness in the films cured at 300 0C in air and at 400 0C in nitrogen suggest that chemical crosslinking is taking place. Ultraviolet irradiation after thermally curing at 100 0C substantially increases the yellowness of the film. The yellowing of the UV irradiated film is indicative of the formation of chemical crosslinks. Table 3.1. Color and Brittleness in Films Cured at Progressively Higher Temperatures. Treatment 100 0C 2000C 3000C 300 0C 400 0C 100 0C+UV Environment air air N2 air N2 air Color light yellow light yellow light yellow dark orange brown yellow Brittleness flexible flexible flexible brittle brittle barely brittle 3.3.2. Thermogravimetric Analysis Thermogravimetric analysis was used to track the presence of solvent as the bulk films are cured. Thermogravimetric scans in Figures 3.2 and 3.3 also indicate vaporization of absorbed water and, at high temperatures, decomposition of the polymer. All of the films lose about 1% of their original weight below 150 0C due to removal of water from the polymer, while the films lose varying amounts of weight, depending upon the cure conditions, between about 150 0 C and 300 0C due to removal of residual solvent. The TGA scans reach a plateau until the temperature exceeds about 400 0C, when the films begin to degrade. 109 In Figure 3.2(a-e), five thermogravimetric scans of Probimide® 412 films are compared. Weight loss (left vertical axis) and derivative of weight loss (right vertical axis) are shown as functions of temperature. Each sample has a different highest curing temperature which is indicated by the vertical marker. None of the films in Figure 3.2 has been UV cured. The amount of residual solvent vaporized between 150 0 C and 3000 C decreases substantially as the cure temperature increases. That is, each cure at a higher temperature progressively removes more solvent from the film. There is no residual solvent in the film cured at 3000 C, Figure 3.2(e). The temperature at which the solvent boils off during thermogravimetric analysis also increases with the increased cure temperature. This can be more clearly seen in the plots of the first derivative of weight loss (with respect to scan time) in Figure 3.2. The temperature at which maximum solvent vaporization occurs increases from about 200 0 C to about 3000 C as the cure temperature of the film is increased from room temperature to 200 0C. The observed distribution of solvent removal temperatures, ranging from under 0 150 C up to about 3000 C, is most like the result of polymer-solvent binding. Loosely bound solvent molecules readily evaporate below the boiling point of the solvent, while tightly bound solvent molecules require much higher temperatures in order to evaporate. As the material is cured, loosely bound species vaporize first, and the distribution of removal temperatures of residual solvent moves to higher temperatures. In Figure. 3.3, thermogravimetric scans of unirradiated and UV irradiated film cured at 100 0 C are compared. Not only is there less solvent in the UV cured film, but also the volatilization of the residual solvent occurs at higher temperatures after UV exposure. It is thus apparent that the UV curing, by crosslinking the polymer chains, affects the solvent removal dynamics of the thick film. As Figures 3.2 and 3.3 show, at around 400 0 C, Probimide® 412 begins to decompose. The decomposition temperature is arbitrarily defined as the temperature at which the time derivative of weight loss equals 3% of the polymer per minute (scanning at 100C/minute). For our Probimide® 412 samples, decompostion is around 5200 C in flowing air. Other authors [10,11] previously report a decomposition temperature of 5270 C in flowing nitrogen. 110 olnu!w / % Iq6!eAA 0 Qt L. E I- % LIB!eM Figure 3.2. Thermogravimetric weight loss and derivative of weight loss vs. temperature for Probimide® 412 thick fims with highest cure temperature: (a) room temperature, 111 elnu!w / % lbIq!eM o Ln lil (D 0 **MO o LM E Lu E I- 0O u, r %Flq.!eM Figure 3.2 continued. (b) 100 0C 112 elnu!u / %l1I!oM 0 0 tO U') 0o L.. EI G) E F- % L16!0M 0 Figure 3.2. continued (c) 140 C 113 elnu!iw / % q6!eM LO ~ C14 6 ~_~_ ___ L ~~IIC -_ ~~I_ -----·I~----------···C·ll ·-- · · · CD 6 0o Q L CIO CL E i-c0 % q6!1eM Figure 3.2. continuted. (d) 2000 C. Derivative curves plotted with two different full scale values, curve (1), 10%/minute, and curve (2), 0.2%/minute. 114 aenu!W / % %LB!aM o 0 (D 0 LO tO 0 4.' a E I- % ,l61!eM Figure 3.2. continued. (e) 3000 C. 115 0 0 (D 0 V0 0 0 =3 -I 0 CL 0E E 0 0 0 t, 0 LO M• ua3Jaad 0 tw .i4l!aM Figure 3.3. Thermogravimetric weight loss vs. temperature for Probimide® 412 thick films baked at 100 0C, UV irradiated or unirradiated. 116 3.3.3. Dynamic Mechanical Analysis The evolution of mechanical properties during the cure of Probimide® 412 bulk films was investigated by dynamic mechanical analysis. The dynamic mechanical spectra at 1 Hz for Probimide® 412 samples progressively treated at 100 0C, 2000 C, 3000 C, and 4000 C are shown in Figure 3.4(a-d) respectively. Young's modulus and dissipation factor, tan8, are shown as functions of temperature. For Figure 3.4, the 3000 C and 400 0 C cures were performed in nitrogen. The DMA scans show both glass transition and secondary transition behavior. The room temperature Young's modulus does not appear to be dependent upon cure conditions. The glassy moduli of our films are about 2.2-2.5 GPa. Other authors report a modulus of 2.6-2.9 GPa for hard baked Probimide® 400 series films [10,13,14]. For the samples cured in air at 100 0C, Figure 3.4(a), and at 200 0C, Figure 3.4(b), Young's modulus decreases by about an order of magnitude at 375 0 C. Tan5 shows a strong maximum at 375 0C, which is the glass transition temperature. Several broad and weak maxima occur near 100 0C and 250 0C. In the film cured in nitrogen at 300 0C, Figure 3.4(c), the glassy modulus decreases slightly at 2800 C and sharply at 375 0 C. Above 4250 C, the modulus exhibits an upturn due to the thermal induction of covalent crosslinks. The dissipation factor, tan5, shows a strong glass transition at 3750 C and a weaker, but distinct, secondary relaxation at 2800 C. In the film cured at 4000 C in nitrogen, Figure 3.4(d), above room temperature, the modulus declines with increasing temperature. There is a secondary relaxation at 280 0C, and a broad relaxation above the 375 0C glass transition temperature seen in the other samples. TanS shows several broad steps. There is no clear glass transition peak seen in this sample. The glass transition temperatures of all of the samples are near 375 0C. However, while the samples treated at 1000 C, 200 0 C, and 300 0 C have sharp glass transitions at 375 0 C, the sample treated at 4000 C has a broad relaxation shifted to higher temperatures. Ree et al. [14] used dynamic mechanical analysis to show that thermal cures from 3500 C400 0C induce chemical crosslinks in Probimide 412. Our results confirm that the 400 0C treatment in nitrogen causes the polymer to crosslink, reinforcing the rubbery modulus and shifting the softening transition to higher temperatures. We also show that the thermal crosslinking does not occur in the films cured at 300 0 C and below. 117 Secondary relaxation also differs among the samples. Ree et al. [14] observed broad and weak P relaxation at low temperatures and attributed the phenomenon to the relaxation of phenyl moieties or to absorbed moisture. Our samples cured at 1000 C and 2000 C have weak and broad secondary relaxations. On the other hand, the samples cured at 3000 C and 4000 C, which have been shown by TGA to be free of solvent, have well defined relaxations at 280 0 C. Ultraviolet irradiation crosslinks the Probimide ® 412 thick films, affecting its relaxation behavior. In Figure 3.5, unirradiated and UV irradiated samples initially cured at 100 0 C are compared. The crosslinking causes an elevation of the glass transition temperature by a few degrees, reinforces the rubbery modulus, and broadens the tan8 peak. Ree et al. [14] observed that UV exposure reinforced the rubbery modulus in samples that had been subsequently hard baked at 350 0 C in nitrogen. They found, however, that thermally induced crosslinks created by subsequent curing at 4000 C dominated over the UV induced crosslinks such that no effect of the UV could be observed. The polymer also crosslinks when it is cured at 3000 C in an oxygen-containing environment. This is demonstrated in Figure 3.6, where the mechanical properties of samples cured at 3000 C in air and nitrogen are compared. While the sample baked in nitrogen has a sharp glass transition at 375 0C, the sample baked in air has a broad glass transition and a highly reinforced rubbery modulus. It is apparent that the 300 0C cure in air has a similar effect on the mechanical properties as does the 400 0C cure in nitrogen. A 4000 C cure in air, however, severely oxidizes and degrades the specimen. 118 9uel O C) aL.a C) (D W E I I (ed) snlnpofl I I I I s,BunoA o-1 Figure 3.4. 1 Hz dynamic Young's modulus and loss factor, tans, vs. temperature for ProbimideO 412 thick films with highest cure temperature: (a) 100 0 C, 119 S9ue 0 c0 0S - - - - co 0 - ---- I I CD 0 -- -- 0 - I --- - -CD .C ,#*a 0Lo 0 00 CD_ o 0 -- -l- I 1~-·1111~·1)~---·· --- -- I · I~-·---~-~----·-- (ed) snlnpol/ s,SunolA Figure 3.4. continued. (b) 200 0 C, I o-i -- I E CD 120 2Uel 0 6 dI 65 SI 0 d 6 1I- v- ew 0 o 0O 0 0 __~ I ,-) -~ ,. ~ ___, - C co0 (ed) snlinpoj I I 1 o-(0 , s,bunoA to'1j Figure 3.4. continued. (c) 300 0 C (nitrogen), - ---- ·---- oCDJ E 0 !-. 121 0 -- I -- - I~--- I -- . " 0 0 0 I'K E CH / _ 11_1 __ I_ I~ -_~_ --~------~-T-LIC I --. mI ii (ed) snlnpo/V s,6unoA 6o-1 Figure 3.4. continued (d)400 0C (nitrogen). -s- 122 @ uo/ U0j 0o tO a\J 00 ) ) ) ) C.) o ::3 -4-0) 1) k-..; f w/ %yJ ( Od) snlnpoqi s Dunok bo Figure 3.5. 1 Hz dynamic Young's modulus and loss factor, tan6, vs. temperature for Probimide® 412 thick films cured at 100 0 C, UV irradiated (dashed line) or unirradiated (solid line). 123 u i 0 0 LO 0 0 0 O 0 00 0 tO 0 0 ,;I 0 0 0 0 E) 0 0o0 E t) HL 0 0 0 LOr"11% 0) (od ) snlnpolAy s,bunol , o07 Figure 3.6. 1 Hz dynamic Young's modulus and loss factor, tan8, vs. temperature for ProbimideO 412 thick films cured at 300 0 C, in air (dashed line) or in nitrogen (solid line). 124 3.3.4. Index of Refraction The index of refraction of the polyimide changes as it is cured. Measurements of the refractive indices (. = 632.8 nm) of spin coated Probimide ® 412 films are shown in Figure 3.7. Average indices obtained for bulk films were quite similar to those obtained for the spin coated films, and thus, only the results for the thin films are presented. The average index is about 1.61-1.62 for samples treated at 1000 C and 2000 C in air, and for the sample treated at 3000 C in nitrogen. However, the index of refraction increases after the films are treated at 3000C in air or at 4000 C in nitrogen. We attribute the large increase in refractive index to thermally induced crosslinking which takes place at 3000 C in air, but at 4000 C in nitrogen. These results are consistent with the observed color changes and with changes in the dynamic mechanical relaxation spectra. Previous work [8] reports an inplane index (%= 1.06 gtm) for Probimide® 400 series films of approximately 1.61, although the authors did not observe an increase in refractive index upon hard curing to 400 0C. Ultraviolet curing also affects the optical properties of the Probimide® 412 film. Figure 3.8 shows the index of refraction of 100 0 C softbaked samples as a function of UV irradiation time. Refractive index increases with UV exposure as a result of crosslinking. Gelation occurs rapidly at first, but slows with larger exposure times. 3.3.5. Fourier Transform Infrared Spectroscopy We used FTIR spectroscopy to determine how thermal and UV curing changes the chemical structure of the polyimide. Figures 3.9, 3.10, and 3.11 show the effects of curing conditions on the FTIR spectra of Probimide® 412. Figure 3.9(a-d) demonstrates the effect of UV exposure, while Figures 3.9 and 3.10 show the effects of thermal cures. Figures 3.9 and 3.11 are scaled to three times the height of the N-C imide peak at 1370 cm- 1. Figure 3.10, which shows only the 2000-4000 cm- 1 region, is scaled to one-third the height of the 1370 cm- 1 peak. Table 3.2 lists the main absorption bands observed and the bond motion tentatively assigned to the vibrational frequency. Several IR peak are of particular interest. Polyimides are well known to absorb in the infrared strongly at 1720 cm -1 and weakly at 1770 cm - 1 due to carbonyl absorption [15]. According to Higuchi et al. [16] weak absorption can be observed at 1680 cm -1 due to benzophenone carbonyl. This was based upon the observation that only benzophenonecontaining polyimides absorb at this frequency [17]. The broad absorption regions observed near 3000 cm -1 and 3500 cm- 1 are also of interest. The absorption near 3000 125 1.65- 1.64- 1.63 - 7/V" 1.62 1.61- A- /$0 I ~ 1 A •- 0 - ml- le - U Y 0 U U Cure Conditions Figure 3.7. Indices of refraction of spin coated Probimide® 412 films as a function of highest cure temperature. 126 1.625 I S 1.620- 6 0 I 1.615 - 120 180 240 300 UV Cure Time (minutes) Figure 3.8. Indices of refraction of spin coated Probimide® 412 films as a function of UV exposure time. 127 cm -1 is attributed to a series of overlapping hydrocarbon stretching bands [18,19], while the broad absorption near 3500 cm- 1 is attributed to the stretching of highly hydrogen bonded hydroxyl groups [18,19]. UV Effects The chemical mechanism of UV crosslinking in inherently photosensitive polyimides has been studied [16,20-22]. UV crosslinking occurs as follows. Hydrogen is first abstracted from an alkyl group (see Scheme I). The resulting free radical attacks the benzophenone carbonyl group (in preference to the imide ring carbonyls), altering the bond to a carbon-hydroxyl group (see Scheme II). R R H 'C 'C' R I: +H hv o II: + H R -- 'C' R R R Probimide® 412 films spun on KBr were softbaked at 100 0C for 15 minutes and cured under ultraviolet. The FTIR spectra at different UV exposure times are shown in Figure 3.9(a-d). Significant absorption band changes that occur with increased UV exposure are observed in Figure 9. First, broad absorption near 3500 cm-1 systematically increases with UV exposure, indicative of an increase in hydroxyl groups. Second, the broad absorption near 3000 cm-1 declines, indicative of a decrease in hydrocarbon bonds. Third, the peak at 1680 cm-1 due to the benzophenone carbonyl group [16] systematically recedes into a shoulder. This is accompanied by a decline in carbonyl absorption at 1720 128 cm -1. We also note the peak at 1770 cm- 1 due to the imide ring carbonyl bonds does not seem to be affected. Thermal Effects The effects of thermal cures on infrared absorption are shown in Figure 3.10 (2000-4000 cm- 1 range) and Figure 3.11 (400-2000 cm -1 range). Changes in the FTIR spectra occur when the polymer is cured at 300 0C in air, Figure 3.10(b,c) and Figure 3.11(b,c), or at 400 0 C in nitrogen, Figure 3.10(d) and Figure 3.11(d). Under these conditions, the IR spectra from 2000 to 4000 cm- 1 in Figure 3.10 show a decrease in absorbance from C-H bonds (near 3000 cm- 1) and an increase in absorbance from O-H groups (near 3500 cm- 1). This suggests that the mechanism of thermally induced crosslinking may be similar to that of the UV induced crosslinking. The IR absorption spectra from 400 to 2000 cm -1 in Figure 3.11 indicate that, as in the UV exposed samples, the carbonyl peak at 1720 cmn1 declines and the benzophenone carbonyl peak at 1680 cm- 1 becomes slightly flatter. We also notice that there is general broadening of all peaks across the finger print region of the spectra in the thermally crosslinked samples. Virtually all of the peaks appear wider and less resolved under conditions of thermal crosslinking. Thus, there are probably many random, nonspecific degradation and crosslinking reactions simultaneously taking place during the thermal cure. The aliphatic-group attack upon carbonyl may be one of many reactions. 3.4. Conclusions Cures above 2000 C are required to remove solvent from thick films of Probimide® 412. However, ultraviolet curing also drives out solvent. Regardless of cure treatment, however, the films contain about 1%absorbed water. Thermogravimetric analysis reveals a decomposition temperature around 5200 C. Probimide® 412 undergoes glass transition softening at about 3750 C and a secondary relaxation at about 280 0 C. Although residual solvent does not appear to affect the primary glass transition, it may induce secondary relaxation at lower temperatures. Ultraviolet curing at X = 364 nm crosslinks the polymer. The crosslinks reinforce the rubbery modulus and broaden the distribution of relaxation times. The UV exposure yellows the polymer and increases its refractive index. The crosslinking mechanism is 129 Table 3.2. IR absorption regions of Probimide ® 412. Wavenumber (cm -1) 3600-3400 3000-2800 1770 1720 1680 1610 1600 1490 1420 1370 1300 1240 1100 850 710 t see references 15-19 Tentativ e Assignmentt O-H C-H C=4D(imide) C=O (imide and benzophenone) C=O (b4 enzophenone) C-C(phenyl) C-C(phenyl) C-C(phenyl) C-C (phenyl) N-(C(imide) C-C-C (t,enzophenone) C-F N-( C(imide) phenyl substitution phenyl substitution s = strong, m = medium, w = weak Strength* w w m s w w w m m s m m m w m 130 0o o 0 o00 E 0 C',~ 0 0) (sl!un AJejl!qJe) eoueqJosqv e!leAIeH Figure 3.9. Relative IR absorbance vs. wavenumber for spin coated Probimide® 412 films softbaked 15 minutes at 100 0 C and exposed to UV. Arrows point to carbonyl (1680, 1720 cm-1), hydrocarbon (2800-3000 cm-1) and hydroxyl (3400-3600 cm -1) bands. (a) 0 minutes (no irradiation). 131 o o C0 O (sl!un A~Jejl!qje) aueqjosqV aA!lelae Figure 3.9 continued. (b)2 minutes, E O T 132 I *L4 O 0 C- E Cl -0 -o -- -- - (sp~un rijejl!qje) a~ueqjosqVO· Figure 3.9 continued. (c) 10 minutes, -- C3 a0 0 J------ (Cs ljejGHt 133 E0 0 o C) CD ( (spun Ajejl~qle) e3ueqjosqV eanejlaj Figure 3.9 continued. (d) 50 minutes. 134 a) 0 C. cu 0 _, C_ Cu 0. a) FL' P a) U) > C-. U, a) .0= Cu dO Cu, ~Cu 4000 3500 3000 2500 2000 Wavenumber (1/cm) Figure 3.10. Relative IR absorbance vs. wavenumber (2000-4000 cm- 1 range) for spin coated Probimide® 412 films with highest cure temperature treatment: (a) 300 0 C (3 hours in nitrogen), (b) 300 0 C (1 hour in air), (c) 300 0 C (3 hours in air), (d) 4000 C (1 hour in nitrogen). 135 0 0 Lt) o 0 0 r E 0 Lu .0 E c~. 00 0, (sl!un AJeja!qJe) eoueqJosqv e!jeletj Figure 3.11. Relative IR absorbance vs. wavenumber (400-2000 cm -1 range) for spin coated Probimide® 412 films with highest cure temperature treatment: (a) 300 0C (3 hours in nitrogen), {arrows point to carbonyl bands at 1680 cm -1 and 1720 cm- 1) 136 o LOl 0 0o gE, r) 42 0 igL C (spun AjeJlqJe) 9ueqJ~osqv (s!un A~e.Il~q~e) eoueq~osqy Figure 3.11. continued. (b) 3000 C (1 hour in air), lcE ar!elaeie 137 = -- c - I---~----~ o O *J 0 o 0O E 0) -o 1m - -=- - - ------------ -- --- (sl!un Ajejl~qje) 9oueqjosqV 8A!jejajl Figure 3.11. continued. (c) 300 0 C (3 hours in air) o 0 (q4 138 II I __1 _·_1_1____1__________ -o O -o O U O L. 00 E Ca -o Lr _ ----- --- -- -~II (sl!un A3ed!q.e) eoueqJosqur - en!ie le). Figure 3.11. continued. (d) 4000 C (1 hour in nitrogen). -I 139 highly specific and occurs by hydrogen abstraction and attack of the free radical at the benzophenone carbonyl. The material also undergoes substantial thermally induced crosslinking which is especially favored in an oxygen containing environment. We found extensive crosslinking to occur when films are baked at 4000 C in nitrogen or at 3000 C in air. A nitrogen environment appears to inhibit crosslinking at 300 0C. The thermal crosslinking process makes the polymer more brittle, reinforces its rubbery modulus, and broadens its distribution of relaxation times. Also, it gives the material a deep yellow to brown color and increases the index of refraction by approximately 20%. Although one possible crosslinking mechanism may be similar to that of the UV process, the thermally induced crosslinking is probably not limited to one mechanism. 3.5. References 1. H. Hiramoto, Mat. Res. Soc. Symp. Proc., 167, 87 (1990). 2. R. Rubner, H. Ahne, E. Kuhn, and G. Kolodziej, Photo. Sci. Eng., 23(5), 303 (1979). 3. J. Pfeifer and 0. Rhode, in Recent Advances in Polyimide Science and Technology, W.D. Weber and M.R. Gupta, Eds., Society of Plastic Engineers, New York, 1987, p. 336. 4. 0. Rhode, P. Smolka, P. Falcigno, and J. Pfeifer, Polym. Eng. Sci., 32(21), 1623 (1992). 5. Japanese patent #5034699, A (1993). 6. J. Cognard, Mol. Cryst. Liq. Cryst. Suppl. Ser., Suppl. 1 (1982). 7. K.K. Chakravorty and C.P. Chien, SPIE Int. Conf. on Adv. in Intercon. and Packaging, 1389, 559 (1990). 8. C.P. Chien and K.K. Chakravorty, SPIE Opical. Thin Films III: New Developments, 1323, 338 (1990). 140 9. R. Mo, T. Maw, A. Roza, K. Stefanisko, and R. Hopla, Proc. 7th Int. IEEE VLSI Multilevel Intercon. Conf., 390 (1990). 10. T. Maw and R. Hopla, MRS Symp. Proc., 203, 71 (1991). 11. T. Maw and R. Hopla, MRS Symp. Proc., 227, 205 (1991). 12. T. Maw, M. Masola, and R. Hopla, Polym. Mat. Sci. Eng., 66, 247 (1992). 13. M. Ree and K. Chen, Polym. Prep., 31(2), 594 (1990). 14. M. Ree, K. Chen, and G. Czornyj, Polym. Eng. Sci., 32(14), 924 (1992). 15. H. Ishida, S. Wellinghoff, E. Baer, and J. Koenig, Macromol., 13, 826 (1980). 16. H. Higuchi, T. Yamashita, K. Horie, and I. Mita, Chem. Mater., 3, 188 (1991). 17. R. Dine-Hart and W. Wright, Die Makromolekulare Chemie, 143, 189 (1971). 18. R. Silverstein and G. Bassler, Spectrometric Identification of Organic Compounds, John Wiley & Sons, New York, 1967. 19. C. Pouchert, The Aldrich Library of Infrared Spectra Edition III, Aldrich Chemical Company, Milwaukee, WI 1981. 20. J. Scaiano, A. Becknell, and R. Small, J. of Photochem. Photobio. A, 44, 99 (1988). 21. A. Lin, V. Sastri, G. Tesoro, A. Reiser, and R. Eachus, Macromol., 21(4), 1165 (1988). 22. J. Scaiano, J. Netto-Ferreira, A. Becknell, and R. Small, Polym. Eng. Sci, 29(14), 942 (1989). 141 Chapter 4. Effect of Cure Conditions on Probimide® 32 Polyamide-imide 4.1. Introduction Polyimides, as discussed in the previous chapters, are important members of the class of high temperature heterocyclic polymer materials [1-3]. They are well known for their outstanding thermal stability, mechanical and electrical properties, and chemical resisitance. Polyimides are used in electronic, display, and aerospace applications. Polyamide-imides [4] are related polymers whose repeat units contain both an imide ring linkage and an amide linkage. Polyamide-imides are less expensive than polyimides, readily processed from solutions, and still have outstanding properties. There is expanding interest in polyamide-imides as engineering plastics [4], as materials for molecular composites [5,6], and as liquid crystal display (LCD) alignment films [7]. In Chapter 1, it is explained that polyimides generally require high temperature treatments which are necessary both for ring closure of the polyamic acid precursor, and for the removal of high boiling point solvents. Studies show [8-16] that the chemical and physical properties of polyamic acid precursor systems change substantially as they are cured, primarily because of the chemical conversion from polyamic acid to polyimide. Unlike most polyimides, polyamide-imides are available in solutions of their fully converted (i.e., fully imidized) form. The fully imidized polyamide-imides systems do not require post-processing heat treatment to effect closure of the imide ring. It is important that the outstanding features, as well as the limitations, of polyamideimides be well understood. It is furthermore crucial to understand these materials from a processing-structure-properties perspective. Since reducing the cure temperature of polyimides may reduce processing cost and time, as well as allow their compatibility with inexpensive low temperature materials, it is interesting to learn how these systems respond to differing thermal treatments. Thus, this work explores the effects of cure temperature and environment on the properties and structure of a preimidized polyamide-imide system. Our goal is to learn how the pre-converted polymer changes physically and chemically during its cure treatment. We report here the processing-structure-property relationships in Probimide® 32 polyaimide-imide as a part of an ongoing evaluation of the material for liquid crystal alignment applications. 142 4.2. Experimental Section 4.2.1. Polymer film processing We studied the material Probimide® 32, a fully imidized polyamide-imide resin produced by OCG Microelectronic Materials, Inc. The polymer is provided in an NMP/xylene solution with 21% solids content by weight. The material has the following chemical repeat unit: + n Bulk films were formed by the technique described in Section 3.2.1 and depicted in Figure 3.1. The Probimide® 32 solution was dropped onto a glass plate and wiped with a metal doctor blade with a 1 mm gap. The wet films were dried to a tack free condition under a dry air current at room temperature overnight. The films were then progressively cured at 100 0C for six hours, then 2000 C for three hours, and then 3000 C for three hours. Thermal cures of 100 0C and 2000 C were performed in air with the polymer on glass, while the 3000 C cures (which we will call "hard cures") were performed on free-standing film in either air or nitrogen environments. Film thicknesses were on the order of 100p.m. All films were amorphous, as the samples exhibited no Bragg scattering peaks when examined by wide angle X-ray scattering. Thin films were formed by spin casting the undiluted Probimide® 32 solution on glass or KBr at 4000 RPM for 30 seconds. The samples were immediately softbaked at 100 0 C for fifteen minutes in air. The samples appeared to be dry after softcuring. The samples then progressively underwent one hour cures at 2000 C and then 3000 C. The 200 0C cures were performed in air, while the 3000 C cures were performed either in air or nitrogen. 143 4.2.2. Measurement techniques A Seiko TG/DTA320 was used for thermogravimetric analysis of the bulk films. Samples were scanned from room temperature to 5750 C at 100C/minute in flowing air (300ml/minute). Dynamic mechanical analysis was performed in tension mode on bulk films using a Seiko DMS200. Samples were scanned under nitrogen at 20C/minute from room temperature to the temperature above Tg at which breakage occurred. Frequencies 1, 2, 5, 10, and 20 Hz were used. Index of refraction at X = 632.8 nm was measured using a Metricon PC-2000 prism coupler. Refractive index was measured by determination of the critical angle. Both inplane nTE and out-of-plane nTM indices were measured. Average refractive index navg was calculated as: navg = (1/3) (2nTE + nTM). Refractive index measurements were taken for the free standing films and the thin films spin coated on glass slides. Fourier transform infrared spectroscopy was performed using a Nicolet 510P. FTIR spectra were taken for the Probimide® 32 thin films spin coated on KBr. Differential scanning calorimetry (DSC) measurements were made for bulk films using a Perkin Elmer DSC-7 calibrated to zinc and indium standards. The scan rate was 200 C per minute. 4.3. Results and Discussion 4.3.1. Coloration and Brittleness The changes in coloration and brittleness in Probimide 32 thick films that occur with subsequent cures are summarized in Table 4.1. The films were considered "brittle" if they were easily broken upon flexing. Probimide® 32 becomes progressively darker as it is cured to higher temperatures, and it is darkest when cured at 3000 C in air. The polymer is brittle after it is cured at only 100 0 C. However, the brittleness recedes when it is cured to 2000C. The film then becomes brittle if cured at 300 0C in air, but does not become brittle if cured at 3000 C in nitrogen. The brittleness and dark coloration of the air cured, hard baked film suggests that abundant chemical crosslinks have formed. (1) 144 Table 1. Color and Brittleness in Films Cured at Progressively Higher Temperatures. Temperature Environment Color Brittleness 100 0C 200 0C 300 0C 3000C air air air N2 yellow brown dark, reddish brown brown very brittle flexible brittle flexible 4.3.2. Thermogravimetric Analysis Thermogravimetric spectra of Probimide® 32 bulk films chart the removal of solvent and water from the films, as well as indicate decomposition of the polymer. Our primary goal is to observe solvent content as a function of cure conditions. We demonstrate here how solvent is progressively removed from the bulk polymer films as they are cured at 100 0C, 2000 C, and 300 0C. Polyimide Solvents Polyimide solvents typically have very high boiling points. N-methylpyrrolidone boils at 202 0 C, while xylene boils at 138 0C-1440 C. However, while solvents unbound to polymer typically evaporate at temperatures below their boiling points, solvents bound to polymer may require higher temperatures in order to vaporize. In the 100C/minute thermogravimetric scan of Fisher Scientific N-methylpyrrolidone in Figure 4.1, all solvent has evaporated by about 190 0C. Probimnide32 In Figure 4.2(a-d), thermogravimetric and derivative scans of Probimide® 32 treated at room temperature, 100 0 C, 2000 C, and 300 0 C, respectively, are compared. The scans depict the loss of water at 800 C, loss of residual solvent at about 2000 C, and decomposition of the polymer above 3500 C. The films lose about 3% of their initial weight below 150 0C due to water removal. All samples treated at or below 2000 C have a large loss in weight in the 150 0 C to 350 0 C region due to vaporization of residual solvent. For the derivative curves, the higher the cure temperature, the higher the temperature at which maximum rate of weight loss of solvent occurs. While the rate of weight loss is maximal at 145 elnu!uw / %jq6!eM 0 0 oL. I-E 0 Figure 4.1. Weight loss and derivative of weight loss vs. temperature for N-methyl pyrrolidone (NMP). 146 elnu!W / % 16I!eM 0 C) L 3 EI CL 0. E C) % I6G!aM Figure 4.2. Weight loss and derivative of weight loss vs. temperature for Probimide® 32 thick films: a.) dried under an air current at room temperature, 147 elnu!W /% lqSi!eM 0 (0 CDj C) I- E 'mm 0 a % tI5!eM Figure 4.2. continued. b.) then baked 6 hours at 100 0C, 148 elnuw/% Lq6i!M 0 (0 lq LO) 0 %%O -e LM CLI E )I- %0/;tt6!eM /0 61RMA a1~V Figure 4.2. continued. c.) then 3 hours at 200 0 C, 149 elnu! / % lqI!eM o 0 0D qzr O E )- 0 CO % 1L6!eM Figure 4.2. continued. d.) then 3 hours at 300 0 C. 0 I%- 150 around 200 0 C in the room temperature cured sample, the rate of weight loss is maximal at almost 300 0 C in the 2000 C cured sample. The peak height decreases as the cure temperature increases. In the 150 0C to 350 0C region, the room temperature cured sample (Figure 4.2(a)) has the largest weight loss, the 100 0C cured sample less (Figure 4.2(b)), and the 200 0C cured sample even less (Figure 4.2(c)). Thus, each cure at a progressively higher temperature removes more solvent from the film. There is apparently no residual solvent remaining in the 300 0C cured film (Figure 4.2(d)). The weight loss curves shows a plateau until the temperature exceeds 4000 C, at which point film degradation begins. As also shown in the previous study of Probimide® 412, there is a distribution of solvent removal temperatures ranging from below 150 0C to above 350 0 C. The 32 films, however, contained more solvent than the 412 films of Chapter 3. The distribution, hence, is even more apparent. Solvent molecules removed at or below the boiling point of the solvent represent loosely bound species that readily vaporize, while molecules that are removed at temperatures far above the boiling point represent species that are physically bound to the polymer. Thus, as the material is cured, low boiling species are removed, and the distribution of removal temperatures of residual solvent species is pushed to higher temperatures. Cure time also affects the residual solvent in the polymer film. Figure 4.3 depicts thermogravimetric scans of Probimide® 32 treated for 1, 6, and 24 hours at 100 0C. As cure time increases, the amount of residual solvent decreases. There is also a small shift in the distribution of solvent removal temperatures between 1 and 6 hours and a larger shift after 24 hours of cure. As Figures 4.2 and 4.3 show, at around 450 0C, decomposition of the polymer begins, as indicated by a large loss in weight. If we arbitrarily define the decomposition temperature as the temperature above 350 0 C at which the time derivative of weight loss equals 3% per minute (scanning at 100C/minute), the decomposition temperature of Probimide® 32 is about 550 0 C in flowing air. The manufacturer reports a decomposition temperature of 494 0 C [17] though they do not report the scan conditions. 151 O (0 C) O t) o 0 O O 4-- E O tO 0 0 O 0 C 0 O (1 0O lua3aJd 14q!aM Figure 4.3. Weight loss and derivative of weight loss vs. temperature for Probimide ® 32 thick films baked at 100 0C for 1, 6, and 24 hours. 152 4.3.3. Dynamic Mechanical Analysis The dynamic mechanical relaxation of Probimide® 32 films provides a wealth of property and structure information. DMA data demonstrate how the glassy regions, the relaxation regions, and the rubbery regions are affected during the cure and how they are dependent upon cure environment. The changing dynamic mechanical spectra can be attributed to changing solvent content and to changing degrees of crosslinking. The DMA spectra for Probimide® 32 are shown in Figure 4.4(a-d) for samples cured at 1000C, 2000C, 3000C in air, and 3000C in nitrogen, respectively. Young's modulus E' and loss factor tan8 are shown as functions of temperature. For brevity, only measurements at 1Hz are shown. The glassy Young's modulus at 250C is affected by the cure conditions of the polymer film. The glassy modulus of film treated at 1000C and 2000C (Figure 4.4(a,b)) is 1.9-2.0 GPa. The glassy modulus of Probimide® 32 film treated at 3000C in nitrogen (Figure 4.4(d)) is somewhat higher, 2.7 GPa, which agrees with reported values of the modulus [17]. It appears that residual solvent in the films cured at 1000C and 2000C plasticizes the films, lowering the glassy modulus. The measured glassy modulus of film treated at 3000C in air (Figure 4.4(c)) is fairly low, 1.3 GPa. This is either because of oxidative damage inflicted upon the polymer during the cure, or because of an inability to make an accurate measurement due to the brittleness of the film. In Figures 4.4(a,b), Young's modulus for samples cured at 1000C and 2000C drops by about two orders of magnitude as a result of polymer relaxation. In the tan8 curves, two relaxations are seen in these samples. The first relaxation in each scan occurs at about the curing temperature, while the second relaxation occurs just below 3000C. The material is highly plasticized by residual solvent, giving the material a low softening temperature. Once the solvent is driven out, the material again relaxes at the glass transition temperature of the unplasticized polymer. In Figures 4.4(c,d), the samples treated at 3000 C, in air or in nitrogen, have only a single relaxation. The glass transition relaxation is centered above 3000C in the air-treated sample. The tan8 peak is shallow, suggesting the sample is highly crosslinked by the 3000 C cure in air. The sample treated under nitrogen at 3000C, on the other hand, has a well defined glass transition relaxation at 3000C. It is thus apparent that the degree of crosslinking in Probimide® 32 is substantially greater when hard cured in air than when hard cured under flowing nitrogen. The rubbery modulus region is also very different in the Probimide® 32 films. The films cured at 1000C and 2000 C have extremely low rubbery moduli, the 2000C cured film 153 having a slightly greater rubbery modulus than the 100 0 C treated film. In the 300 0 C nitrogen-treated material, the rubbery Young's modulus reaches a minimum value of 0.16 GPa shortly after relaxation. This rubbery modulus is about ten times greater than that observed in the 100 0 C and 200 0 C treated films. Apparently, the 3000C cure in nitrogen not only increases the glass transition temperature, but also provides reinforcement to the rubbery state, probably as a result of backbone crosslinks that form during the hard cure. The rubbery modulus of the film cured 300 0 C in nitrogen slowly increases with temperature as the high temperatures induce more non-oxidative crosslinks in the polymer. In the film cured at 3000 C in air, no rubbery region is clearly observed because gel network has already formed before the polymer relaxes. Instead, the modulus slowly declines and does not appear to flatten entirely before the sample breaks. In a polyimide that is not preconverted, the modulus and glass transition temperature increase as the polymer is cured. This is not surprising, as the conversion of polyamic acid to polyimide increases the stiffness of the chemical backbone. This study shows, however, that even in a fully converted system, residual solvent can also have a marked effect on the mechanical properties of the material. The study also demonstrates that the extent of interchain covalent bonding strongly impacts the mechanical behavior of the polyamide-imide. 4.3.4. Index of Refraction The optical properties of Probimide® 32 are also cure-dependent. They change with cure temperature and cure environment. The indices of refraction of thin films of Probimide® 32 spin coated on glass are presented in Figure 4.5(a,b). The average indices of refraction of the bulk films were identical to those of the thin films. Average indices of refraction of spin coated films are plotted against highest cure temperature in air in Figure 4.5(a). There is a general increase in refractive index with increasing cure temperature. These changes coincide with the observed changes in coloration described in Table 1. The increase in index is associated with increased density and/or electronic polarizability due to a combination of several factors: loss of solvent, crosslinking, and increased interchain interaction. Average indices of films cured at 300 0C are shown in Figure 4.5(b) as a function of 3000 C cure conditions. The films hard baked films under nitrogen have slightly lower average refractive indices than do those treated in air. This suggests that the crosslinking reaction at 3000 C is faster in air than in nitrogen. 154 2Uejl. CO o; d o oo 0 0 0 a0 O L. o (1) E 0C SZIL. ('• 0N 0 1Q) 0 (ed) sninpoins, ,Aor i tuno Figure 4.4. Young's modulus and dissipation factor, tan8, for Probimide® 32 thick films: a.) cured 6 hours at 100 0C, 155 9ueA o 6 d d 6o6 do 6 0 0 CD0 0 rafCD, M C)) CD 0 L 0g 2 I-~Z E 0 T W) Co (ed) snlnpoi s,6 unoA Figure 4.4. continued. b.) then 3 hours at 200 0 C, QD 15o tf% 156 0D 0U. D 9- Ir ·I, +1 C* CD CDJ om # E 0 1 0 I (8d) sfIfPOV II SUno U Bo (ed) sninpoyy s,Bunol( 6o-I Figure 4.4. continued. c.) then 3 hours at 300'C in either air or, C 157 ue.l VCO ltft0Ok a U ~7 r; I-, 0 0D 0 0 CD) C) E 1a v a, v. m w M (ed) snlnpoN s, unoA Bo-1 Figure 4.4. continued. d.) nitrogen. I ^ 158 1 II 75E 1.70- U U 1.65- U 1.60 100 200 300 Highest Cure Temperature (oC) Figure 4.5. a.) Indices of refraction of spin coated Probimide ® 32 films vs. highest cure temperature in air. Cure cycle was 15 minutes at 100 0 C, 1 hour at 2000 C, 1 hour at 3000 C. 159 1.75 ICI 1.70- ~ II_ ~I_ ~ I , //-- 1.65- /0 1.60 II Nitrogen (1 hr.) -*- r l4 | Nitrogen (3 hrs.) Cure Conditions b.) Refractive indices for different cure conditions at 300 0 C. Air (1 hr.) I 160 4.3.5. Fourier Transform Infrared Spectroscopy Fourier transform infrared spectroscopy is a powerful method of examining the progress of a polyimide cure. This is especially true in polyamic acid precursor materials, which undergo significant changes in their infrared spectra as they are converted to polyimides. Our fully preconverted polyamide-imide, on the other hand, gives us an interesting opportunity to observe the effects of solvent removal and oxidative crosslinking in the absence of the amic acid to imide conversion. We demonstrate here how the FTIR spectrum of Probimide® 32 evolves during curing. The study below suggests a technique for monitoring the cure in fully converted polyimide and related systems. Solvent Spectra N-methylpyrrolidone (NMP) and xylene are the two solvents used in the Probimide® 32 system. Their chemical structures are shown below: CH3 I CH3 I F•I CH3 NMP xylene FTIR spectra of these compounds can be found in most infrared spectra atlases [18]. The main IR absorption regions and our peak assignments are shown in Table 4.2. The three chemicals in the Probimide® 32 system, the polyamide-imide polymer, NMP, and xylene have a number of chemical units in common. For example, both the polyamide-imide and NMP contain an amide unit, the polymer a secondary amide, and the NMP a tertiary amide. The carbonyl bond of the amide group in both molecules contributes an extremely strong infrared absorption band at about 1680 cm- 1. Also, the xylene and the polyamide-imide contain aromatic bonds, which have a number of IR absorption bands, which are listed in Table 4.2. One key spectral peak, due to carboncarbon skeletal stretching of the aromatic ring, appears at around 1500 cm-1. There can be multiple bands in this region at slightly different absorption frequencies depending upon the ring substituents. In addition, all three substances have carbon-hydrogen bonds. These bonds typically have broad absorption in the locality of 3000 cm- 1. 161 Table 4.2. IR absorption regions of NMP and xylene. Compound Peak Location or Approx. Region (cm - 1) Tentative Assignment? Peak Strength* NMP 2800-3000 v(C- H) s 1680 v(C = O) s 1380-1500 6(C- H), 5(N-CH 3 ) s 1300 v(C-N) s 1110 v(C -N) m 980 3(ring) m 2900-3100 v(C - H) s 1380-1650 v(C = C) s 1000-1050 aryl 8(C - H) m 790, 760, 740, 690 8 ring substitution bands s xylene see references 18,19 s = strong, m = medium, w = weak 162 Probimide 32 Spectra Fourier transform infrared spectroscopy of spin coated thin films of Probimide® 32 on KBr, shown in Figures 4.6-4.9, revealed two processes that take place in the system upon curing. The first process is the systematic removal of solvent. The second is crosslinking and degradation at 300 0 C in air. Assignments of the main absorbance bands [18,19] in the polyamide-imide polymer are shown in Table 4.3. The presence and removal of residual solvent is apparent in the "finger print" region of the FTIR scans. Figure 4.6(a-f) shows the 400-2000 cm- 1 region of Probimide® 32 samples treated for 10 minutes at 20 0C intervals between 100 0C and 200 0 C. As the sample is cured from 100 0C to 200 0 C (Figure 4.6(a-f)), a number of peaks decline or disappear. These changes are consistent with removal of solvent. The most important peak is associated with the amide carbonyl bond, which is present in both the polyamide-imide and the NMP solvent. The tertiary amide in the NMP and the secondary amide in the polyamide-imide both absorb at about 1680 cm-1. This strong absorption peak gradually declines in Figure 4.6 as NMP is removed from the polymer by treating it to progressively higher temperatures. Observation of the strong reflection at 1510 cm-1 due to aromatic skeletal stretching also proves quite interesting. As the polymer is treated to higher temperatures, an overlapping side peak systematically recedes. The best explanation for this occurrence is that NMP or xylene, either of which could contribute to this side peak, is systematically removed. When the solvent is eliminated, a shoulder to the peak at 1510 cm -1 remains, reflecting the fact that there are two differing aromatic substitution environments in the polyamide-imide. As the polymer is cured, we also observe the gradual disappearance of a shoulder peak near 1300 cm- 1, decline of a peak at around 1110 cm- 1, and disappearance of a small peak at about 980 cm- 1. All three of these changes are consistent with the gradual removal of NMP from the system. We also see an increase in the band 1770 cm- 1 associated with in-phase stretching of the imide ring carbonyl. However, there is no obvious explanation for this. The 2000-4000 cm- 1 region is shown in Figure 4.7. Two marked changes are observed. First, there is decline in absorption in a broad range around 3000 cm- 1. This indicates a decrease hydrocarbon bonds, and thus, is consistent with the removal of solvent. Second, there is gradual shifting of the N-H amide peak from about 3300 cm-1 to almost 3400 cm-1. It is typical for the location of this band in amide compounds to exist at 163 Table 4.3. Partial list of IR absorbance bands in Probimide® 32 polyamide-imide. Peak Location or ~ ~ L~ Tentative Approx. Region (cm- 1) --- I~-3100-3400 (broad) -- ~--- Assignmentt v(N-H) Peak Strength* 3000-3100 (broad) v(C-H) 1770 v(C=O) (imide) w ---------- w --------- m ~--- v(C=O) (imide) -~--~------ 5 ------ 1720 "--- v(C=O) (amide) -I--- v(C=C) (phenyl) L I-~..I s-r II 1680 ~--- 1600 -- 1510 ----·---- S II 1410 c m --- -- 1380 ~I--- 1320 ~I~ 720 _ ~ v(C=C) (phenyl) -------~ -~I- v(C=C) (phenyl) I----~--. v(N-C) (imide) v(N-C) (amide) ------- 8 phenyl substitution _C _ _ see references 18,19 s = strong, m = medium, w = weak m -- 5 _ ~ -LI~I~-·C--~-C-- m 164 a lower wavenumber in the solid state than in the solution state as a result of increased hydrogen bonding [19]. A shift to a higher wavenumber as solvent is removed suggests that the polymer forms stronger hydrogen bonds with the NMP than it does with itself. In Figure 4.8(a-d), 400-2000 cm -1 FTIR scans region of Probimide® 32 treated at 100 0 C, 2000 C, and 3000 C in air, and 3000 C in nitrogen, respectively, are shown. The effects of solvent removal described above are clearly seen in these scans as the films are treated above 100 0C. Also, we note that the peak at 1770 cm-1 corresponding to in-phase stretching of the imide carbonyl increases as the polymer is heated from 100 0 C to 2000 C, but returns to its original height after treating at 3000 C in either air or nitrogen. Again, there does not seem to be an obvious explanation for this. The 2000-4000 cm- 1 region is shown in Figure 4.9. Reduced absorption near 3000 cm-1, as well as shifting of the N-H peak from 3300 to 3400 cm-1, both a result of solvent removal, are again apparent as the material is treated above the softbake temperature. Previously, we showed that a 300 0 C treatment in air initiated substantial chain crosslinking, while the same thermal treatment in nitrogen did not. There are noticeable differences the FTIR spectra of Probimide 32 films treated at 300 0 C in air, Figures 4.8(c) and 4.9(c), and at 300 0 C in nitrogen, Figures 4.8(d) and 4.9(d). These changes suggest possible mechanisms for the oxidative crosslinking reaction. The most apparent difference is a reduction in the air-treated sample in C=C phenyl stretching near 1510 cm-1. This is accompanied by a drop in C-H stretching near 3000 cm- 1. In fact, two distinct hydrocarbon peaks near 2900 cm- 1 are nearly eliminated. The amide C=O peak at 1680 cm-1 appears to become nearly a shoulder of the imide C=O peak at 1720 cm- 1. Also, there is an increase in the C=C phenyl vibration at 1600 cm- 1. Kuroda and Mita [20] also observed a large drop in the 1510 cm-1 phenyl vibration, accompanied by an increase in the 1600 cm- 1 phenyl vibration, in their studies of thermooxidative degradation of a polyimide. They interpreted the decline in the 1510 cm-1 peak as the cleavage of 4-CH 2 bonds. The authors speculate that the increase in the vibration at 1600 cm- 1 might be associated with the formation of diphenyl crosslinks, as suggested by Dine-Hart et al. [21,22], though Kuroda and Mita [20] do not suggest this to be a primary path of crosslinking in their material. Dine-Hart et al. [21,22], in their investigations of the oxidative curing of PMDAODA polyimide, concluded that the main crosslinking reaction appeared to be the intermolecular coupling of phenyl rings in the diamine segments of the main chains. This occurred by the formation of phenolic groups which subsequently react. The phenolic bonds were formed either by direct dehydrogenation of the phenyl rings, as shown in Scheme I, or by scissioning of aryl-ether bonds. 165 [1O] OH H -HO20 Scheme I The reduction in the 1510 cm - 1 phenyl vibration, increase in the 1600 cm- 1 phenyl vibration, and reduction in C-H stretching near 3000 cm- 1 in the Probimide 32 polyamideimide might be associated with the formation of the crosslinks shown in Scheme I. The diphenyl crosslinks must occur in the diamine portion of the polyamide-imide repeat unit, as the 1510 cm- 1 vibration is associated with 1,4-C 6H4 groups [23], while the 1600 cm -1 band can be attributed to more highly substituted phenyl groups [23]. The inter-chain crosslinks in Scheme I might occur in the conjunction with the intra-chain breakage of 1,4phenyl-phenyl bonds in the diamine, as shown in Scheme II. The radicals in Scheme II may subsequently produce random bond reformations. Scheme II 166 Broadbelt et al. [24] reviews the various reactions that may occur on the amide linkage of an aromatic amide model compound as the substance undergoes elevated temperature treatments. A reaction mechanism that would be consistent with an increase in molecular weight would be a [1,3] sigmatropic shift (see Scheme III) followed by random radical recombination reactions. It is also likely that many of the other reactions could simultaneously take place, including various bond homolyses of the amide bond followed by random radical recombinations which do not result in a net increase in molecular weight. If the sigmatropic shift and random crosslinking at the amide bond were to occur in the polyamide-imide during its 3000 C treatment in air, we would expect to see decreases in amide C=O stretching at 1680 cm- 1 and in N-H stretching in the 3100-3400 cm-1 region. These changes appear to have occurred in the FTIR spectra for the 3000 C air treatment in Figures 4.8(c) and 4.9(c)). 0 I "O-N' 0 + OH0 + OH I + H20 Scheme III The above data suggest that cure can be monitored in the polyamide-imide system by FTIR. This could be an especially valuable tool if, for example, a new processing scheme and cure cycle were to be tested for solvent removal effectiveness. The most apparent change in the FTIR spectrum as solvent is eliminated is the decline of the amide carbonyl peak at 1680 cm -1. In Figure 4.10, the peak height at 1680 cm - 1 relative to that at 1370 cm -1 is plotted against highest cure temperature for FTIR data shown in Figures 4.6 167 and 4.8. The ratio declines steadily with increasing cure temperature. Because the spin coated films are thin, we expect the films to be nearly free of solvent after the one hour cure at 2000C. The drop in peak height after the 3000C cure could be due to either further removal of high boiling point solvent species or to non-oxidative crosslinking at the amide carbonyl. The films appear to be solvent free when the relative absorption ratio is about 0.6-0.7. 4.3.6. Differential Scanning Calorimetry The calorimetric relaxation response of Probimide @ 32 was studied as a function of cure temperature and environment in order to gain further insight into the processes that take place during curing. DSC scans of Probimide® 32 bulk films treated at 1000C, 200 0C, and 3000C (air and nitrogen) are shown in Figure 4.11 (a-d), respectively. Several physical phenomena can be observed in these scans: the vaporization of water, the vaporization of solvent, and transition from the glassy to rubbery state. Vaporization of water contained by the polymer, also observed in the TGA scans in Figures 4.2 and 4.3, is seen in all DSC scans in Figure 4.11. It appears as a broad endothermic response near 1000C. Vaporization of residual solvent is observed in the 1000C treated sample, Figure 4.10(a), as indicated by the large endothermic response seen in the scan between 1500C and 300 0C. It is possible that the spikes observed are a consequence of instrument noise caused by the violent vaporization occurring in the encapsulated sample. Any glass transitions that might have occurred in the 1000C treated sample are hidden by the strong vaporization response. Solvent vaporization is not clearly indicated in any other sample. In the 2000C treated sample, Figure 4.11(b), a glass transition is seen at about 2000C, as indicated by the step increase in heat capacity. A weak response at about 3000C is also seen. These data are in good agreement with the DMA results shown in Figure 4(b). In both the 3000C air-cured and nitrogen-cured samples, glass transitions are seen at 3000C. The transition in the air-treated sample, Figure 4.11(c), is broader and slightly shifted to a higher temperature, as compared to the transition in the nitrogen-treated film, Figure 4.11(d). This is consistent with the DMA scans shown in Figure 4.4(c,d). However, in the DSC scans, a weak step in the heat capacity is seen at about 3300C, where no comparable change in mechanical modulus was observed in the DMA data. 168 o 0 0 0 E 0O 2, (sl!un AJeJl!qje) eoueqJosqv eA!,eleH Figure 4.6. Relative absorbance vs. wavenumber (400-2000 cm - 1) for spin coated Probimide® 32 films. Arrow points to 1680 cm - 1 carbonyl peak: a.) 10 minutes at 100 0C, 169 o 0 0O E cc LO 0 0l OC o o (sp!un Ajejl!qJe) eoueqjosqV eA!lelae Figure 4.6. continued. b.) then 10 minutes at 120 0C, 170 LO o E 0O ow eoueqJosqVeAlec (sun ) AJl!qJ (sl!un AWeJ4!qw) eoueq~osq~ Figure 4.6. continued. c.) then 10 minutes at 140 0 C, ea!4eje~ -- 171 0 0 0 0 E 0. I,, E ,, 0U) 0n 0 Cun ueqosq el (sp~un Aawl!qje) eoueqaosqV8Alj~ejO Figure 4.6. continued. d.) then 10 minutes at 160 0C, 172 o 0 LO 0 0 E .0 0) (Q (sl!un Ajejlqae ) 0oueqaosq-V O~llelati Figure 4.6. continued. e.) then 10 minutes at 180 0 C, 173 o 0 LO 0 E E LO OCN (sl!un AjeJl!qJe) eoueqJosqV eA!IeBll Figure 4.6. continued. f.) then 10 minutes at 2000 C. 174 0 i a E L. MW ~ 0 0 (sl!un •Je~j!qJe) aoueqJosqy VA!lelGJ Figure 4.7. Relative absorbance vs. wavenumber (2000-4000 cm-1) for spin coated Probimide® 32 films: a.) baked 10 minutes at 100 0C, b.) then 10 minutes at 120 0 C, c.) then 10 minutes at 140 0 C, d.) then 10 minutes at 160 0C, e.) then 10 minutes at 180 0 C, f.) then 10 minutes at 200 0C. 175 Lu C) .0 E C) (sl!un Ajej!qje);eoueqJosqv GAPIele1 Figure 4.8. Relative absorbance vs. wavenumber (400-2000 cm - 1) for spin coated Probimide® 32 films: a.) 15 minutes at 100 0 C (arrow points to 1680 cm -I carbonyl peak), 176 o 0 LO E 0) CD (sl!un AjeaJ!qJe) eoueqjosqVoq eA!•elt E Figure 4.8. continued. b.) then 1 hour at 200 0 C (arrow points to 1680 cm - 1 carbonyl peak), 177 I- -------- ~~I --- -- ~--------- -D0LO c · ~O I 0 -0 0 E 0 nn E LO C) t-o j I- 1-----~--(sl!un AJeJ.!qJe) eoueqjosqVi --- I -D ea!Iejea Figure 4.8. continued. c.) then 1 hour at 300 0 C in either air or (arrow points to 1510 cm-1 C=C peak), 178 o 0 U) 0 E_. t* (sl!un Ajejl!qje) eoueqJosqV eanlCelEi Figure 4.8. continued. d.) nitrogen (arrow points to 1510 cm - 1 C=C peak), 179 0D 0 cD 0 0 0 LnE 0.._ LM. 0 oE M LO (sp!un Ajejl!qJe) aoueqjosqV a^JelaOE Figure 4.9. Relative absorbance vs. wavenumber (2000-4000 cm -1) for spin coated Probimide® 32 films: a.) baked 15 minutes at 100 0 C, b.) then 1 hour at 2000 C, c.) then 1 hour at 300 0C in either air or, d.) nitrogen. 180 1.5 1.0 0.5 II 0.0 II Highest Cure Temperature (oC) Figure 4.10. Ratio of relative absorbance of peak at 1680 cm - 1 to that of the 1370 cm - 1 vs. highest cure temperature. 181 181 0 Lt Cn 0 0 CD) 0 Ln C14 0 CD CJ C*4LC 4) E C) 0 Lt 0 0 0 In qf 0 < - I 0 0 0 (I/M) MOI- leaH o!WJuLaqIo0PU3 Figure 4.11. Endothermic heat flow vs. temperature for Probimide® 32 thick films: a.) baked 6 hours at 100 0C, 182 o 0 0 0 0 CO CD CDO U) 0 0 CD oQ)c CL E CDC CD =I- 0 0V 0) 3- o(51M) MJ 1 MOl_-I •,•eH o = Ig• oI 0 0 LOn O!LWeqlopu3 (DIM) MOld IeaH o!WueLHyopu3 Figure 4.11. continued. b.) then 3 hours at 2000 C, 183 o 0 LOl cv, 0 0t•M LO oT o C C CD 4) -ca 1-0 0 0 u, C I 1 d (D1M) MOlI- leaH 3!luJeqLopu3 Figure 4.11. continued. c.) then 3 hours at 300 0C in either air or, 0 0 ·I 184 o LO 0o cy, 0 Cn 0 CD 0 LN 0 1) (N o L. E i-r) C r - ) Mq Ci (5/AA) Mo13 lean o!ruJlelopu3 Figure 4.11. continued. d.) nitrogen. o 185 4.4. Conclusions Although the polyamide-imide Probimide ® 32 undergoes no thermal imidization reaction during curing, the structure and properties of the polymer system are dependent upon its curing conditions. The effects of progressive cure conditions on the properties and physiochemistry of the polymer have been studied by several characterization methods. When bulk films of Probimide® 32 are treated at only 1000 C, a great deal of solvent remains. The residual solvent affects the properties and deteriorates the thermal stability of the material. The volume of residual solvent is reduced and the properties and thermal stability improved when the polymer is cured to 2000 C. Samples treated at 300 0 C were solvent free and were highly thermally stable. However, substantial oxidative crosslinking occurs if the polymer is cured at 3000 C in air rather than in an inert environment such as nitrogen. 4.5. References 1. C.E. Sroog, MacromolecularReviews, Vol. II, J. Polym. Sci. (1976). 2. S.D. Senturia, in Polymersfor High Technology - Electronicsand Photonics, M.J. Bowden and S.R. Turner, Eds., American Chemical Society, Washington, D.C., 1987, p. 428. 3. J.W. Verbicky, Jr., "Polyimides" in Encylopedia of Polymer Science and Engineering, J.I. Kroschwitz, Ed., John Wiley and Sons, Inc., New York, 1987. 4. H.R. Slater, Advanced Materials and Processes, 1, 19 (1995). 5. R.J. Karcha and R.S. Porter, Polymer, 33(22), 4866 (1992). 6. S. Palsule and J.M.G. Cowie, Polym. Bull., 33(2), 241 (1994). 7. D.C. Rich, E.K. Sichel, and P. Cebe, MRS Fall Meeting 1995, in press. 8. M.H. Kailani, C.S.P. Sung, and S.J. Huang, Macromol., 25, 3751 (1992). 186 9. P.R. Dickinson and C.S.P. Sung, Macromol., 25, 3758 (1992). 10. E. Pyun, R.J. Mathisen, and C.S.P. Sung, Macromol., 22, 1174 (1989). 11. J.R. Ojeda, J. Mobley, and D.C. Martin, J. Polym. Sci. B, 32, 559 (1995). 12. J. Jou and P. Huang, Macromol., 24, 3796 (1991). 13. C.A. Pryde, SPE ANTEC '90, 439 (1990). 14. M.-J. Brekner and C. Feger, J. Polym. Sci. A, 25, 2005 (1987). 15. E.D. Wachsman and C.W. Frank, Polymer, 29, 1191 (1988). 16. R.W. Snyder, B. Thomson, B. Bartges, D. Czerniawski, and P.C. Painter, Macromol., 22, 4166 (1989). 17. OCG Microelectronic Materials, Inc., Probimide 32 product bulletin. 18. E.g., C.J. Pouchert, The Aldrich Library of Infrared Spectra Edition III, Aldrich Chemical Company, Inc., Milwaukee, WI (1981). 19. R. Silverstein and G. Bassler, Spectrometric Identification of Organic Compounds, John Wiley & Sons, New York (1967). 20. S. Kuroda and I. Mita, Eur. Polym. J., 25(6), 611 (1989). 21. R. Dine-Hart, D. Parker, and W. Wright, Br. Polym. J., 3, 222 (1971). 22. R. Dine-Hart, D. Parker, and W. Wright, Br. Polym. J., 3, 226 (1971). 23. H. Ishida, S. Wellinghoff, E. Baer, and J. Koenig, Macromol., 13, 826 (1980). 24. L. Broadbelt, S. Dziennnik, and M. Klein, Polym. Degr. Stab., 44, 137 (1994). 187 Chapter 5. Alignment of Liquid Crystals on Brushed Polymer Layers. 5.1. Introduction Despite the tremendous industrial importance of, and scientific interest in, the surface alignment of liquid crystals on polymer films, it is still a poorly understood phenomenon. Berreman and others [1-3] have argued that the liquid crystals align along microgrooves created by the rubbing. The irregularity in size and location of these scratches [4-8] in buffed polymer films, however, might cast some doubt upon this theory. On the other hand, researchers have shown that the non-rub formation of microgroove patterns can in fact produce alignment of nematic liquid crystals [9,10]. For example, Kawata et al. [10] have shown that liquid crystal alignment can be achieved using fine microgroove surface patterns with pitches less than 2gm produced by photolithography. More recently, authors have suggested that molecular orientation of the main chains produced by the brushing is responsible for alignment [8,11-19]. For example, Aoyama et al. [11] have shown that liquid crystals can be aligned on polymer films that have been stretched but not scratched. In addition, authors have observed that the direction of alignment is not always identical to the direction of the scratches. For example, while the grooves in rubbed polystyrene films run parallel to the brushing, the liquid crystal alignment is perpendicular to the brush direction [6-8]. This suggests that orientation of the phenyl side groups, which are approximately perpendicular to the main chain, might be responsible for alignment in these films. It seems reasonable that if the physical groove pattern were the only factor responsible for alignment, then nearly any grooved polymer film should be able to align liquid crystals and the alignment should always occur in the same direction. Geary et al. [13] have suggested that high crystallinity in polymer films is associated with the polymer's ability to align liquid crystals. Russell [18] reports, using a surface-sensitive X-ray reflectivity technique, that brushed polyimide films develop an oriented crystalline structure at the surface. On the other hand, polyimides, though widely used for liquid crystal alignment, are not widely known for being crystalline. The BPDAPDA polyimide studied by Russell [19] is one that is known to develop high degrees of crystallinity [20], which is an unusual characteristic for polyimides. Though polyimides are typically amorphous, it is possible that crystalline structures could form at the polymer 188 surfaces. Factor et al. [21], for example, report crystalline structure on the top 90A of PMDA-ODA polyimide films. Polyimides with long alkyl side groups and bulky fluorinated groups are frequently used for liquid crystal alignment. These alignment materials are known to generate large pretilt bias angles in the liquid crystal layer [22-27]. In addition photocrosslinkable polyimides are commercially available for liquid crystal alignment applications [28]. These polymers allow the alignment films to be patterned by photolithography. It is unlikely that any of these polyimides could crystallize due to the bulky side groups or to the photocrosslinked main chains. Hence, this casts substantial doubt that crystallinity could be a necessary feature for liquid crystal alignment. In conclusion, the surface alignment of liquid crystals on polymer layers is not well understood despite the significant amount of work that has been done. Conventional thin film and surface evaluation techniques do not seem to be generating highly insightful information regarding the phenomenon. In this chapter, the alignment of liquid crystals on Probimide 32 polyamide-imide and other polymer layers is demonstrated and discussed. A optical transmission technique for quantitatively evaluating parallel alignment is presented. A highly novel technique for investigating thermophysical transitions in surface layers is then demonstrated. 5.2. Experimental Section The following section describes the experimental techniques used for producing and evaluating liquid crystal alignment cells. It was necessary to use laboratory techniques that optimally reflected the goals of, and were realistic to, the resources available. Thus, extensive consultations with LCD researchers at the 3M Corporation [29] were helpful in developing the experimental approach described in this section. The goals of these experiments were to: 1. Obtain data that are physical meaningful and have real scientific merit. 2. Present proof of new concept research on the laboratory scale. 3. Focus on the materials science of the liquid crystal surface alignment phenomenon, rather than on LCD device technology in general. 189 Many decisions were made during the course of this project in order to serve and focus upon the above goals. For example, it was decided to build cells in a university laboratory rather than in an industrial clean room. A hand brushing technique was used rather than a commercial buffing wheel for the induction of alignment. Merck E7 liquid crystal mixture, which is mainly used in research rather than in commercial displays, was selected as the nematic material. Parallel nematic alignment was chosen as the geometry for the evaluation of alignment. The techniques described below are optimal methods with respect to the purposes of this research. 5.2.1. Construction of liquid crystal alignment cells Preparation of substrates The following procedure was adopted for preparing glass substrates for spin coating. The purpose of the following cleaning procedure was to remove grease, dirt, and dust particles from the glass. This was desired so that the substrate coverage and smoothness of the polymer coating could be optimized. Poor substrate preparation resulted in poor polymer coatings. 1. Glass microscope slides (25 x 75 x 1 mm) were cut approximately in half by mechanically etching the slide down the middle along its width. The etched slides were then gently broken in half. 2. The slides were blown clean of particles and dust with an air jet stream. 3. The slides were soaked for 15-20 minutes in isopropyl rubbing alcohol for removal of organic contaminates. 4. The slides were rinsed of the alcohol by soaking for 15-20 minutes in distilled or deionized water. 5. The slides were cleaned in an ultrasonic bath with detergent for 20 minutes. 6. The slides were rinsed in distilled or deionized water for 15-20 minutes. 190 7. The slides were rinsed again in an ultrasonic bath of distilled or deionized water. It was found that this dual rinse procedure was necessary for thorough removal of the detergent. Residual detergent was found to have a significant adverse affect on coating quality. 8. The wet slides were blown dry with an air jet stream. 9. The slides were stored in a PTFE, low-dust sample holder which held twenty glass slides. Spin coating procedure The following general procedure was adopted for applying polymer alignment films to the cleaned glass substrates. I specifically thank Prof. Anne Mayes for use of the Headway Research spin coater. 1. A dilute solution of the polymer was prepared. The primary polymer of interest, Probimide 32 polyamide-imide, was obtained from OCG Microelectronic Materials, Inc., as 21 wt% solids and was diluted to approximately 3 wt% solids with N-methyl pyrrolidone (NMP). For example, to 2.5 mL of Probimide 32 original solution, 15 mL NMP were added. 2. The glass substrates were blown clean of dust with an air jet just prior to spin coating. 3. The glass substrates were prewet with solvent, which was spin coated at 4000 RPM for 30 seconds. For Probimide 32 coatings, the substrates were prewet with NMP. 4. The polymer solution was spin coated onto the prewet glass substrates at 4000 RPM for 30 seconds. 5. The films were softbaked about 15-20 minutes at approximately 110 0C at atmospheric pressure on a hot plate or on a temperature-controlled metal block in a National Appliance Oven. The films appeared dry after the softbake. The resulting Probimide 32 polyamide-imide films were approximately 450A thick, as measured by a Dektak 8000 profilometer. 191 Alignment layer buffing procedure The following technique was used to brush the polymer films for liquid crystal alignment. The procedure is depicted in Figure 5.1. The substrates were temporarily adhered with two-sided tape in a locked position to a glass plate, which was in turn, securely taped to a horizontal bench top. A rayon cloth was wrapped around a cylindrical bar which weighed approximately 1 lb. The bar was dragged across the polymer film by hand exactly 25 times such that the vertical force on the film was equal to that of the weight of the bar. Although this technique is not as controllable as a commercial brushing wheel, researchers at 3M Corporation believe this technique to be highly adequate in producing reasonably consistent alignment films [28]. Buffing cloth Alignment layer surface Figure 5.1. Alignment film brushing technique. Liquid crystal alignment cell building procedure Liquid crystal cells were constructed by the following procedure, which is depicted in Figure 5.2. 1. Two alignment films on glass were prepared and brushed as described above. 2. PF-70 7gpm diameter microrod spacers obtained from Nippon Electric Glass were mixed into a globule of Norland Optical Adhesive 61 UV-curable epoxy (Norland 192 Products, Inc., New Brunswick, NJ 08902). The mixture was mixed fresh prior to cell building. A tiny clump of the spacer rods, approximately 0.03 mg, was added to around 3 drops of the epoxy. This was enough mixture to make about 4 cells. 3. The epoxy/microrod mixture was placed near the edges of one substrate parallel to the brush direction. 4. The two slides were pressed firmly together with the alignment films on the inside of the cell and the brushing directions parallel. 5. The epoxy was exposed through the top glass substrate to i-line (? = 365 nm) ultraviolet light under a Spectroline BIB-150B black light UV lamp (average area normalized energy delivered to glass approximately 2.5 mW/cm 2 ) for about 3.5 minutes. The cell was about 10" from the bulb. The energy required to cure the UV sensitive epoxy was many orders of magnitude less than what might affect the polymer films. 6. The cell gaps were filled by capillary action with Merck E7 liquid crystals. The E7 liquid crystals are low-cost monomeric nematic liquid crystals used primarily in laboratory research. The chemical composition of the liquid crystals is proprietary, though Cognard [30] reports from chromatographic analysis the materials to be the mixture of molecules shown in Table 5.1. The physical properties [31] of the E7 liquid crystals are listed in Table 5.2. The liquid crystals were filled parallel to the direction of alignment, a technique which appeared to minimize the cell defects. 7. The cells were maintained at about 800 C on a temperature-controlled metal block in a National Appliance oven for approximately one-half hour and then cooled. This treatment was sufficient to clear the liquid crystals and minimize defects. 8. The cells were sealed with Devcon 5 Minute epoxy so that the liquid crystals could not seep out of the cells. 193 (a) COAT TWO ALIGNMENT FILMS ONTO GLASS; APPLY BRUSH TECHNIQUE (b) CONSTRUCT CELL WITH GAP 711 SPACER RODS + EPOXY (c) FILL GAP WITH NEMATIC LIQUID CRYSTALS SPACERS & EPOXY CI Z' i I ADHERED GLASS SLIDES -. m MERCK E7 NEMATIC LIQUID CRYSTALS (d)MAINTAIN FILLED CELL ABOVE THE LC CLEARING POINT I -- I T . .w (e) SEAL COMPLETED CELL Figure 5.2. Cell construction procedure. (a) Alignment film formation, (b) Empty cell construction, (c) Cell filling, (d) Cell annealing, and (e) Cell sealing. 194 Table 5.1. Composition of Merck E7 nematic liquid crystals from chromatographic analysis [30]. I _ _ 51% _~ __ 1~1 ~_ _ I_ _ NC ~ // ~C 25% 8% I~ __ _ NC-- NC _I 1(~1_ _ _~ ~C I~ 16% _ C5Hj 1 ~ NC I_ __~_ I_ ~__ __ 7 _ H15 ~ICI _ -Cs 8H,7 C5H1 _ 195 Table 5.2. Physical properties of Merck E7 nematic liquid crystals [31]. Melting Point Liquid Crystal Clearing Point Extraordinary Refractive Index at 589 nm Ordinary Refractive Index at 589 nm Birefringence Viscosity (200C) -100C 60.50C 1.746 1.522 0.224 39 mm 2/s 5.2.2. Evaluation of alignment cells Oualitative evaluation of alignment The liquid crystal alignment in the cells was judged qualitatively by illuminating and rotating the cells between two crossed polarizers. The observations made for aligned and unaligned cells are summarized in Figure 5.3. This technique is an excellent first order test of an alignment film's ability to orient the liquid crystals macroscopically. Aligned cells exhibited macroscopic optical birefringence, with four extinctions occurring 900 apart upon rotation when the alignment direction is parallel to either polarizer axis (Figure 5.3(a)). Light is transmitted when the cell is rotated off one of these orientations. The maximum transmitted light intensity occurs when the nematic director is 450 to the polarizers' axes (Figure 5.3(b)). In contrast, unaligned liquid crystal cells were bright (i.e., they evidenced no extinction of light) for all rotations of the cell between crossed polars (Figure 5.3(c)). This is because the unaligned liquid crystals form arbitrary nematic phases with randomly varying nematic directors. The birefringence is microscopic but not macroscopic. 196 A (a) LP ALIGNMENT A (b) tP (c) A I L, NO ALIGNMENT Figure 5.3. Qualitative observations of liquid crystal alignment cells between crossed polarizers. For aligned cells, a.) the light is extinguished when the alignment director is parallel to either polarizer, and b.) maximum intensity is transmitted when the alignment director is 450 with respect to the polarizers. For unaligned cells, c.) the cell appears bright for all rotations. 197 Measurement of optical transmission The optical transmission of the liquid crystal cells between a crossed polarizer (P) and analyzer (A) was measured using the configuration shown in Figure 5.4. The laser is a Spectra-Physics Model 159 He-Ne laser (, = 6328A). Phase sensitive detection was achieved by use of a Stanford Research Systems SR540 chopper and EG&G Model 5209 lock-in amplifier. A diagram of the connections is shown in Figure 5.5, and the settings for the lock-in amplifier are listed in Table 5.3. A Newport 818-SL detector with 883-SL attenuator and Model 835 power meter were used to detect transmitted intensity. The time constant for intensity measurements was 1 second. A chopping speed of 100 Hz was used. The polarizers had an extinction ratio (intensity A//P / intensity AiP) of about 105. The intensity transmitted by the crossed polarizers was negligible compared to the lowest cell transmission measurement. Measured voltage was linear with light intensity for the range studied. This is proven in Figures 5.6 and 5.7. First, a series of calibrated neutral density filters were placed in front of the detector with no other components in the optical train. The measured relative intensity is plotted in Figure 5.6 against expected relative intensity, which was calculated according the extent of calibrated neutral density filtration. Figure 5.6 indicates a linear relationship between real and measured intensity. The data is plotted on a linear scale in Figure 5.6(a) and on a log-log scale in Figure 5.6(b) so that both data approaching one and data approaching zero can be seen. Next, a polarizer and analyzer were placed in the beam. The transmitted intensity was measured as the analyzer was rotated with respect to polarizer. Expected relative intensity was calculated according to the theoretical relationship: I/E2 = cos 2 X (1) where I/E 2 is relative intensity and X is the angle between the polarizer and analyzer. Measured relative intensity is plotted against expected relative intensity in Figure 5.7. The data are plotted on a linear scale in Figure 5.7(a) and on a log-log scale in Figure 5.7(b) so that both data approaching one and data approaching zero can be seen. 198 SAMPLE CHOPPER P A LASER -- - RI Figure 5.4. Optical transmission apparatus with phase sensitive detection. 199 LOCK-IN AMPLIFIER [REFERENCE AC IN] [ANALOG OUTPUT] - OPTIC AL POWE R METE R [REFERENCE M ODE f] CHOPPER CONTROLLER III [CHOPPER HEAD CABLE] [DETECTOR] DETECTOR HEAD 0 CHOPPER HEAD Figure 5.5. Instrument connections for phase sensitive detection. Port labels are stated in brackets where appropriate. 200 Table 5.3. Settings for the EG&G 5909 lock-in amplifier used for phase senstivitve detection of chopped light. SENSITIVITY Input from voltage source (i.e. power meter) at A port and press A key Adjust sensitivity scale with up and down keys as appropriate FILTERS Use Bandpass BP filter mode. Use TRACK to track the reference chopping frequency. Set LINE REJECT such that neither F nor 2F filters are lit. TUNING Hit red AUTO key, then AUTO PHASE key, and wait for red AUTO light to turn off. This procedure automatically determines the phase adjustment for maximum output signal. The adjustment is necessary because the external electronics are not necessarily in phase. Use +90' to shift the phase adjustment by 900, twice to shift by 1800 REFERENCE Connect chopper signal at AC IN port INT (internal reference) should not be lit, since an external reference is being used 2F should not be lit, as this would double the reference frequency OUTPUT Adjust TIME CONSTANT as desired The higher the time constant, the less the noise, but the longer it take to produce data 10 sec is a good time constant to begin with Use SLOPE 6dB, though this is not crucial in most cases DYN RES - NORM DISPLAY - %FS.SIGNAL (% of full scale sensitivity setting of output signal) OFFSET & EXPAND - keep unlit 201 (a) * 0.8- S S 0.60 a: 0 0.40 0.20 p 0.0 I . 0.0 * * 0.2 0.4 0.6 0.8 Expected Relative Intensity 1.0 (b) 0 m .01 " .001 * I .0001 .0001 0 .001 Expected .01 .1 1 Relative Intensity Figure 5.6. Measured relative intensity vs. expected relative intensity for phase-sensitive detection of neutral density filtered light. Expected relative intensity is calculated from the extent of calibrated neutral density filtration. a.) linear scale, and b.) log-log scale. 202 1.0 h - 0.8 - ftt 0.6 0.4 0.2 a~ - 00 < X < 900 900< X< 1800 - a - on m -- L VI1 v.v 0.0 I a I I 0.2 0.4 0.6 0.8 1.0 Expected Relative Intensity (b) I .1 a 00 < X < 900 .01 1 900< x < 1800 ft .001 - .0001 2i - .0001 . . ..... .001 .01 .1 1 Expected Relative Intensity Figure 5.7. Measured relative intensity vs. expected relative intensity for phase-sensitive detection of light diminished by two consecutive polarizers with their polarizing axes lying at an angle X. Expected relative intensity is given by cos 2X. a.) linear scale, and b.) loglog scale. 203 The optical transmission intensities of the liquid crystal alignment cells were measured between the crossed polarizer and analyzer. To calculate the alignment parameter A=I(0 0)/I(45 0) discussed in Chapter 1, both the intensities I(00), brushing axis parallel to the polarizer, and 1(450), brushing axis 450 to the polarizer were measured. Approximately 12 measurements were taken for each orientation, i.e., 00 or 450, of each sample cell. The measurements were taken for sample spots distributed throughout the cell. Highest and lowest values of the data sets were discarded and average values were calculated from the remaining measurements. The experimental error is discussed in Section 5.4. The optical transmission of the liquid crystal cells was measured while adjusting for very small (up to 30) Maguin twist distortions [32] in the cells that appear to have been introduced during cell construction. Such distortions rotate the axis of polarization of the light, but do not represent defects in the surface alignment of the liquid crystals. The analyzer and cells were adjusted such that for the 00 state, the polarizer was parallel to the first LC surface director and the analyzer was perpendicular to the second LC surface director. This is shown in Figure 5.8. The adjustment should counteract the affect of the twist distortion. Failing to adjusting for twist distortions in the 00 extinction position can have a serious affect on the accuracy of the data. The small twist distortions should not have a significant affect on the intensity at the 450 orientation, as can be shown by equation (25) of Chapter 1, and it does not matter therefore whether an adjustment is made for this position. Twist distortions can also introduce ellipticity in the light passing through even if the components are adjusted as shown in Figure 5.8. However, one can show that for small twist angles <30, by making this adjustment, the ellipticity of the transmitted light due to twisting is very small, and that cell can be reasonably approximated as a parallel-aligned layer. According to Gooch and Tarry [33], after making the adjustment shown in Figure 5.8, the relative transmission (T) due to elliptically polarized light is given by: T = [sin 2 (0 (1+u2) 1/2 ] / [1+u 2 ] (2) where 0 is the twist angle and u is given by: u = tihAn/6k (3) where h is the liquid crystal layer thickness, An is the birefringence, and X is the 204 wavelength of light in free space. If we assume h=7.m, An-0.224, X=6328A, and for the worst case scenario, 0=30, the calculated value for T is approximately 4.5x10 -5 , which is more than two orders of magnitude below the lowest measured value of cell transmission. Hence, it is reasonable to conclude that ellipticity due to twist distortion is not significant. P.N Figure 5.8. Adjustment made for a cell with small degree of twist distortion 0. The first liquid crystal surface director N is oriented parallel to polarizer P. The second surface director is N'. The analyzer is adjusted by 0 from A to A' such that A' is perpendicular to N'. 205 5.3. Results and Discussion 5.3.1. Brushing induced alignment in Probimide 32 polyamide-imide The processing, structure, and properties of Probimide 32 polyamide-imide were previously discussed in Chapter 4. In this section, it is shown that this polymer can be used as an alignment layer for nematic liquid crystal. Alignment films of Probimide 32 and liquid crystal parallel alignment cells were produced by the techniques described in Section 5.2. The brushing method was used to induce the liquid crystal aligning properties in the polyamide-imide. Qualitatively, it was apparent that the liquid crystal cells produced by this method were strongly aligned along the brushing axis. The alignment of the liquid crystal cells was evaluated by use of the laser optics method described in Section 5.2. A brush-aligned cell was placed between the crossed polarizer and analyzer. The azimuthal angle of the alignment director with respect to the polarizer was varied. The transmitted intensity is plotted in Figure 5.9(a) as a function of the azimuthal angle of the alignment director. The intensity axis is normalized by the transmitted intensity measured with the alignment director at 450 with respect to the' polarizer. For ideal alignment, we expect the transmitted intensity to follow the expression: I(4)/I(450) = sin 2 (2)) (4) where 0 is the angle of the alignment axis with respect to the polarizer and I(4)/I(450) is the transmitted intensity normalized by that at 0=450. The theoretical relationship is plotted in Figure 1.13 in Chapter 1. It is apparent that the measured relationship shown in Figure 5.9(a) closely approximates the theoretical relationship in Figure 1.13. Thus, it is clear that there is a strong degree of alignment in the brush aligned cell. On the other hand, the intensity at 0', 900, and 1800 is not zero as theoretically predicted, indicating that the alignment of the liquid crystals is not perfect. The ratio of minimum to maximum intensity is on the order of 10-3. The optical transmission of a liquid crystal control cell in which there was no brushing treatment to the alignment film is shown in Figure 5.9(b). Qualitatively, no alignment was observed in this cell. The cell was placed between the crossed polarizer and analyzer of the laser optical transmission apparatus. The length axis of the cell (which would have been the brush axis had the alignment films been brushed) was selected as the axis of reference. The cell was rotated through 1800 and relative intensity was plotted 206 (a) 1 .01 .Ol • .001 I .0001 45 0 135 90 Angle (degrees) 180 (b) C I~- --- ~--~--~ 01 .001 0.001 .0001 - 0 45 90 135 Angle (degrees) I 180 Figure 5.9. a.) Intensity / intensity at 450 vs. rotation angle for a nematic liquid crystal cell aligned with brushed Probimide 32 polyamide-imide layers. The x-axis is the azimuth of the alignment director with respect to the polarizer. b.) Relative intensity vs. rotation angle for nematic liquid crystal cell with untreated Probimide 32 polyamide-imide layers. The xaxis represents the azimuth of an arbitrary reference axis with respect to the polarizer. 207 against the angle between the polarizer axis and the reference axis. It is apparent that, unlike the aligned cell, there is no systematic change in the transmitted intensity as the sample is rotated. This indicates that there is no macroscopic alignment in this sample. For all aligned liquid crystal cells produced with brushed polyamide-imide layers, the average ratio of intensity at 00 to intensity at 450 is approximately 1.04 x 10-3 with a standard deviation of 2.8 x 10-4 (27%). This ratio, which is referred to as the A parameter in Chapter 1, is three orders of magnitude lower than unity, the value of the A parameter which should occur in the case of no macroscopic alignment. Hence, it is clear that there is very strong and significant alignment in these specimens. The effective angle of misalignment (D,defined by: A - sin 2 (2i) (5) is approximately 0.920 for the brushed polyamide-imide aligned cells. Since 4=00 for an ideally aligned specimen and •=45 0 for an unaligned specimen, it is quite apparent that the misalignment of the liquid crystals with respect to the alignment director is very small. 5.3.2. Brushing-induced alignment in other polymers During the course of these investigations, it was desired to find other polymers which effectively align the E7 nematic liquid crystals by the rubbing technique. A brief discussion of the results of this search is worthwhile. In Table 5.4, polymers which were used as alignment films are listed. Their success in parallel-aligning the liquid crystals in the brush direction is stated in the second column of Table 5.4. Some characteristics of the polymeric materials are listed in the remaining columns. The polymers in Table 5.4 were said to have a "rigid backbone" if the repeat unit contained ring structures or bulky side groups. The polymers were said to have "strong intermolecular forces" if the repeat unit contained highly dipolar moieties, such as those involved in hydrogen bonding or charge transfer complexation. Calculated effective angles of misalignment for cells produced with various polymer layers are shown in Figure 5.10. Two characteristics which appeared not to be necessary for aligning the E7 liquid crystals were backbone rigidity and aromatic content. Since aromatic polyimides have both these characteristics, one might have speculated that these characteristics may be necessary for, or associated with, liquid crystal aligning properties. However, this does not appear to be the case. For example, poly(vinyl alcohol), a non-aromatic vinyl polymer, was an effective alignment layer material, while polycarbonate, a rigid aromatic polymer, was not. 208 The polymers with strong intermolecular forces, on the other hand, appear to be strong aligners of the E7 liquid crystals. Polyimides are, for example, known to exhibit strong intermolecular charge transfer interactions. The polyamic acid, though weaker in charge transfer, contains both carboxylic acid and amide groups, which are strongly hydrogen bonding moieties. The polyamide-imide is a hybrid of the polyimide and the polyamide and should also exhibit strong intermolecular forces. Poly(vinyl alcohol) contains strongly hydrogen bonding hydroxyl groups. The poly(phenylene ether sulfone) molecule contains highly electron withdrawing sulfone groups, which should engender strong permanent dipoles in the main chain. All of these polymers have a yellowish appearance in the solid film state, which might suggest a helpful rule of thumb in selecting a good alignment polymer. Table 5.4. List of polymers which were examined as liquid crystal aligning agents and some of their characteristics. Polymer Probimide 32 polyamide-imide Aligns Rigid Aromatic Strong Inter- E7 LCs? Backbone? Groups? molecular Forces? yes yes yes yes Color yellow (PAI) PMDA-APB polyamic acid yes yes yes yes yellow (PAA) PMDA-APB yes yes yes polyimide (PI) poly(vinyl yes . yes no no yellow yes alcohol) (PVA) poly(methyl methacrylate) yellow no no no no none (PMMA) polycarbonate no yes yes no yes yes yes yes none (PC), poly(phenylene ether sulfone) (PPES) yellow 209 7 4035 rn - 30- VU 25201510- i 50* 07ý4 M!2=. cnsmn PI PAA PAl PPES ""7"" PVA PMMA PC Polymer Figure 5.10. Values of D, the effective angle of misalignment, for liquid crystal cells produced with alignment films of brushed PMDA-APB polyimide (PI), PMDA-APB poly(amic acid) (PAA), polyamide-imide (PAI), poly(phenylene ether sulfone) (PPES), poly(vinyl alcohol) (PVA), poly(methyl methacrylate) (PMMA), and polycarbonate (PC). All cells but those produced with PMMA and PC appear to have very strong macroscopic alignment. 5.3.3. FTIR Spectra of Probimide 32 polyamide-imide alignment films FTIR spectra of Probimide 32 films are displayed and discussed in Chapter 4. The films in Chapter 4, however, were spin coated from 2lwt% solids and were approximately 4.m thick. The polyamide-imide films used in this chapter for liquid crystal alignment cells, on the other hand, were coated from 3wt% solids and are approximately two orders of magnitude thinner. The FTIR study in Chapter 4 showed that solvent was systematically removed from the 4pm films as the material was heated from 100 0C to 2000 C. This was primarily evidenced by the significant change in peak height of the 1680 cm - 1 amide carbonyl vibration relative to the 1370 cm-1 imide N-C vibration. The 1680 cm-1 peak declined as 210 NMP solvent was driven from the film. The ratio of the 1680 cm-1 band to the 1370 cm- 1 settled near 0.6-0.7 at point of full solvent removal. The solvent dynamics are significantly different, however, in the thin alignment layer films. Figure 5.11 shows the FTIR spectrum of a thin Probimide 32 alignment layer film on KBr which was cured by only the 110 0 C soft bake treatment. The 1680 cm- 1 vibration appears to be mitigated as compared to the solvent-containing films in Chapter 4. The ratio of the peak height of the 1680 cm-1 band to that of the 1370 cm- 1 is about 0.63, which is consistent with a solvent free film. Thus, this study indicates that the solvent in the Probimide 32 preimidized polyamide-imide resin can be removed rapidly at low temperatures in very thin films (approximately 450A). Brushing the polyamide-imide film induces orientation of the polymer chains in the direction of the brushing. This is consistent with what other researchers [8,11-19] have found in brushed alignment films of other polymeric materials. The orientation is clearly indicated by the dichroic FTIR spectra of brushed films in Figure 5.12(a,b). The polyamide-imide films have been brushed, then scanned in the FTIR after placing a gold wire grid IR polarizer in the path of the beam. For Figure 5.12(a), the IR is parallel to the direction of brushing, and for Figure 5.12(b), the IR is perpendicular to the direction of brushing. It is important here to notice the relative ratios of the 1370 cm- 1 imide N-C peak to that of the 1720 cm- 1 imide C=O peak. These values are listed in Table 5.5. The 1370 cm- 1 band represents vibration parallel to the main chain, while the 1720 cm-1 represent vibration approximately perpendicular to the main chain. The ratio of the heights of these peaks increases approximately 37% from 0.63 to 0.86 as the brushed polymer film is rotated from the perpendicular to the parallel position with respect to the IR polarization axis. 5.3.4. Atomic Force Microscopy of brushed polamide-imide films Atomic force microscopy (AFM) images are shown in Figure 5.13 for brushed and unbrushed Probimide 32 polyamide-imide alignment layer films. For the brushed material, Figure 5.13(a), there appear to be grooves running through the film parallel to the brush direction. These grooves are not present in the unrubbed sample, Figure 5.13(b). These relatively large scratches observed in the brushed material appear consistent with what other researchers [4-8] have observed in brushed polymer layers. Seo et al. [5,7], however, observed grooves structures in rubbed polystyrene and alkyl-branched polyimides, but not in a brushed straight chain polyimides, which they observed to have only a rough, bumpy surface. 211 I 0 -o a 0 -oo - . 1L - 0 .0 2 0) -. 0 1o 0 (sp!un AjeiJ!qJe) eoueqJosqv a^!elal: Figure 5.11. Relative absorbance vs. wavenumber for Probimide 32 polyamide-imide coated from 3wt% solids content and softbaked at 110 0C, with final film thickness approximately 450A. 212 0 0 0 o0 r - E 0 U E I 0 rC ) 0 0 0 0 0 (sl!un AjeJl!qJe) eoueqJosqv eAn!el'e Figure 1.:z. Relative absorbance vs. wavenumber for a brushed Probimide 32 polyamideimide thin film. a.) Incident infrared electric field polarized parallel to the brushing direction. 213 o i -o o - I o 0 -o 0 I I (sF!un AiJeJ!qje) eoueqjosqV ea!ielden Figure 5.12. continued. b.) Incident infrared electric field polarized perpendicular to the brushing direction. 214 zC 0 $~*::~I C I, 0 Ln 0Cý U,> 0u :: l -0 of polyamide-imide films. Figure 5.13. Atomic force microscopy (AFM) images brushed. 215 0 II fn -0iLn 0 -N) N V Figure 5.13. continued. b.) unbrushed 216 Table 5.5. Relative ratios of the heights of the imide N-C (1370 cm- 1) peaks to the imide C=O (1720 cm- 1) peaks for brushed Probimide 32 layers. Scan conditions Absorbance Ratio (1370 / 1720 cm- 1) No brush, unpolarized IR IR // Brushing IR I Brushing 0.76 0.63 0.86 5.3.5. In situ thermal stability of liquid crystal alignment on brushed Probimide 32 polyamide-imide layers. The surface alignment of the E7 liquid crystals on brushed Probimide 32 alignment layers was stable to very high temperatures. A completely constructed cell was annealed for half an hour at progressively higher temperatures in 20 0C intervals from 800 C to 180 0C. The cell was cooled after each annealing and the alignment parameter A = 1(00)/1(450) was measured. Treatment above 180 0C damaged the epoxy, thus ruining the cell. The quality of alignment is plotted against annealing temperature in Figure 5.14. The data indicate that the liquid crystals remain strongly aligned throughout the cure cycle. -1 * *0 .001 .0001 • 60 Il . 80 " . 100 " . 120 * " . 140 0 " I, 160 -I - 180 200 Temperature (oC) Figure 5.14. 1(00)/1(450) vs. temperature for a polyamide-imide aligned liquid crystal cell undergoing an annealing cycle of systematically increasing temperature treatments. 217 5.3.6. "Liquid crystal alignment relaxation" - a new technique for measuring thermal transition temperatures in surface layers. The technique in the previous section for measuring the in situ thermal stability is disadvantageous in that it is limited by the chemical thermal stability of the cell epoxy and the liquid crystals themselves. The following technique was employed to measure the thermal stability of only the polymer alignment films: 1. Coat and cure many polymer alignment layers. 2. Apply brushing treatment. 3. Cure the brushed alignment films at an array of temperatures (with at least two films per treatment temperature). Hold each alignment film at a single temperature for 30 minutes, then cool. 4. Build liquid crystal cells from the post-brush cured alignment films. 5. Measure alignment strength A = I(00)/I(450). 6. Plot alignment strength 1(00)/1(450) vs. post-brush cure temperature. Probimide 32 polyvamide-imide For cells constructed with Probimide 32 polyamide-imide alignment films, the alignment strength is plotted against post-brush annealing temperature in Figure 5.15. There is a sharp break in the alignment near 2700C as 1(00)/1(450) changes from approximately 10-3 to approximately 100. For post-brush temperatures 3000 C and above, 1(00)/1(450) remains around one, indicating that the alignment is completely destroyed. The 300 0C "alignment relaxation temperature" correlates with the glass transition temperature for solvent-free Probimide 32 found in Chapter 4. The change that occurs in the polymer surface layer can be visualized as shown in Figure 5.16. The original coated surface begins as an isotropic random coil in the plane of the film. The brushing process orients the polymer chains in the direction of brushing. This orientation process is associated with the layer's ability to align liquid crystals in the brushing direction. When the material is treated above its glass transition temperature, however, the oriented structure 218 relaxes and the surface's ability to align liquid crystals is irreversibly lost. The alignment with this polymer layer, thus, appears to be associated with the frozen-in ordering of amorphouschains which relax at Tg. The relaxation of the aligning ability of the polymer layer suggests a new technique for measuring the thermophysical transitions in polymer surfaces. It also suggests a way in which processing / structure / property studies of liquid crystal alignment surface layers can be performed. For example, it was observed in Chapter 4 that residual solvent in Probimide 32 bulk films substantially reduced the mechanical softening temperature. The single glass transition temperature observed for the polyamide-imide alignment films near 300 0 C suggests, in agreement with FTIR results in Section 5.3.3, that the 110 0C treated alignment films are essentially free of solvent. This technique is not a direct measure of the bulk Tg such as that obtained by DSC or DMA. This technique measures the ability of the surface to perform a function, i.e., align nematic liquid crystals. Hence, the relaxation technique measures only the thermophysical properties of the material responsible for aligning the liquid crystals. Authors [18] have suggested that only the first 10-60 nm is responsible for the alignment. It is therefore unclear at this point whether this functional Tg is that of the surface monolayer or of the first several nanometers. 1U .** tn o OO .1 1 .01 .001 * 1 .0001 p 50 100 0 * * 5 150 I 200 250 I 300 350 400 450 Temperature (oC) Figure 5.15. 1(00)/1(450) vs. post-brush cure temperature for liquid crystal alignment cells produced with Probimide 32 polyamide-imide alignment films. 219 (a) Brush (left to right) (b) (c) Figure 5.16. Depiction of the structure of the alignment layer films during alignment relaxation process of the polyamide-imide. The amorphous chains begin randomly coiled in the plane of the film (a). They are oriented by the brushing process (b). The orientation is reversed by annealing above Tg (c). 220 In Chapter 4, it was found that a 3000 C curing treatment in air crosslinked the polyamide-imide bulk films, increasing the glass transition temperature. To determine whether this effect can be observed in surface alignment layers, the alignment relaxation methodology was employed. Polyamide-imide alignment films were baked at 300 0 C in air for 30 minutes. This treatment was expected to induce chemical crosslinks in the material and increase its thermal stability. The films were then brushed to produce liquid crystal alignment. Before building the alignment cells, the films were treated at an array of postbrush curing temperatures in order to see what temperature treatment destroyed the alignment. The alignment parameter 1(00)/1(450) is plotted against post-brush curing temperature in Figure 5.17. The data in Figure 5.17 are compared to the data obtained with no pre-brush 3000 C air treatment, i.e., the data in Figure 5.15. Clearly, the 3000 C air treatment has increased the thermal stability of the films. The alignment material can withstand higher temperature treatments before its aligning capacity is destroyed because the material is chemically crosslinked prior to brushing. IU 00 10 0o O 0 .01 0 0o 0 .001 .0001 I 50 100 I 150 "' I 200 I 250 I 300 350 400 450 Temperature (oC) Figure 5.17. I(00)/I(450) vs. post-brush cure temperature for liquid crystal alignment cells produced with brushed polyamide-imide that were either softbaked (filled circles) or cured at 3000C in air (open circles) prior to brushing. 221 Poly(phenylene ether sulfone) Poly(phenylene ether sulfone) was found to have strong liquid crystal aligning properties. It was selected as a candidate for alignment based on its highly polar backbone and its yellowish color. The chemical repeat unit of this polymer is shown below: The bulk glass transition temperature is 2260 C [34]. The material was obtained as pellets from Scientific Polymer Products and was dissolved to 3wt% solids in N-methyl pyrrolidone. The material was spin coated to form films for liquid crystal alignment cells. The alignment relaxation technique was employed for the poly(phenylene ether sulfone) films. The relaxation "scan" obtained is shown in Figure 5.18. Destruction of the alignment begins near 210 0 C and is complete near 230 0C, as frozen-in, oriented amorphous chains are relaxed above Tg. 0 S 0 0 .0011 S 0 0 .0001 100 200 300 400 Temperature (oC) Figure 5.18. 1(00)/1(450) vs. post-brush cure temperature for liquid crystal alignment cells produced with brushed poly(phenylene ether sulfone) alignment films. 222 PMDA-APB Polvimide and the Effect of Imidization The alignment relaxation of PMDA-APB polyimide films was investigated. The polyamic acid precursor solution was obtained from the NASA Langley Research Center as 15wt% solids in dimethylacetamide (DMAc). The polymer solution was diluted in DMAc to 5wt% solids and spin coated onto glass slides for liquid crystal cell construction. The alignment film thicknesses were about 1500oA as measured by a Dektak 8000 profilometer. The chemical repeat unit of the fully imidized polyimide is shown below. St. Clair [35] previously reports a 2210 C glass transition temperature for this polymer. The alignment relaxation technique was employed for the softbaked films and for films baked at 2750C for 1 hour prior to brushing. The 1(00)/I(450) alignment parameter is plotted against post-brush cure temperature in Figure 5.19. The relaxation temperature is apparently quite low, approximately 170 0C, in the softbaked films, Figure 5.19(a), because the material has not yet been converted from polyamic acid to polyimide. The glass transition temperature is much higher in the 275 0 C pre-brush treated material, Figure 5.19(b). Tg is around 2500 C, which is only slightly higher than the previously reported 2210 C [35]. Clearly, the imidization reaction has increased the thermal stability of the aligning properties of the material. A series of PMDA-APB films were also pre-brush cured at 2210C for 30 minutes. The alignment relaxation scan is shown in Figure 5.19(c) The thermal stability limit appears to be below 250 0 C. However, there appears to be a reversal of the alignment destruction slightly above this temperature. This reversal might be associated with the completion of imidization of the material. 223 10 _____~_~_____ _____ _1_1_-~---·--·-----·^·---------- ~I __ _ __ _ __ a 1 S0O oo .1 0 .01 .001 .0001 I .001 II_ ,-100 lOO 3 200 300 400 Temperature (OC) Figure 5.19. 1(00)/1(450) vs. post-brush cure temperature for liquid crystal alignment cells produced with brushed PMDA-APB alignment films. a.) PMDA-APB poly(amic acid) softbaked only. 224 b 0 •g • o .01 i-o 0. *0 **0 .001 .0001 a • 0 0 I 100 100 aa " . ! 200 200 a0 I (°C) 300 300 Temperature Temperature (oC) Figure 5.19. continued. b.) PMDA-APB hard baked 2750 C 1 hour. 400 400 225 10; C 0 0 kn .11 B o [] .01 .001 0 • 0 "' o o " .0 13 3 .0001 0 100 200 300 Temperature (oC) Figure 5.19. continued. c.) PMDA-APB cured at 221 0C for 30 minutes. 400 226 Poly(vinyl alcohol) Poly(vinyl alcohol) has strong liquid crystal aligning properties. PVA is also known to be a highly crystalline polymer. Based on that fact, it seems less likely that the orientation of amorphous chains could be associated with alignment on PVA layers. 115,000 molecular weight PVA was dissolved at around 800 C as 3 wt% solids content in water. The solution was spin coated onto glass as films for liquid crystal alignment cells. The relaxation of the alignment is shown in Figure 5.20. There appears to be a transition from aligned to unaligned which begins at around 200 0C and appears to be complete at around 2250 C. This "thermoalignment" transition is consistent with the melting temperature previously reported by Cebe [36]. Thus, it appears that for PVA, the alignment is associated with the orientation of crystalline material. As the crystals melt, the orientation introduced by the brushing process is destroyed. This data suggests that alignment of liquid crystals on brushed polymer layers is a "phase-neutral" process in the sense that it does not matter whether the polymeric film is crystalline or amorphous. -M m 1 0 .1 0 0 * 0 0 .001 0 .0001 I I~ 100 I 200 300 400 Temperature (OC) Figure 5.20. 1(00)/1(450) vs. post-brush cure temperature for liquid crystal alignment cells produced with poly(vinyl alcohol) alignment layers. 227 5.4. Experimental error Experimental error was managed by several techniques, some of which have already been discussed. First, systematic error due to twist distortions was minimized by use of the adjustment technique described previously in Section 5.2.2. Error due to variations in cell thickness from sample to sample was minimized by use of the alignment parameter A=I(0 0 )/I(450 ) which should not be a function of cell thickness. Still, statistical error can arise from variation in both the alignment strength and cell thickness from spot to spot within the sample. This error was minimized by a two-fold strategy. First, many data points were taken for each cell as previously explained in Section 5.2.2. Second, many cells were constructed throughout the course of the research, as is evident by the many data points in Chapters 5 and 6. For I(00) and 1(450) measurements of individual cells, the standard deviations of the means were on the order of 50% or less. In order to calculate the alignment parameter A=I(00 )/I(450 ) for an individual cell, two mean intensities were divided. The propagated standard deviation, Sw, for this mathematical manipulation of the data is given by [37]: Sw2 = (1/y2) Sx2 +(x2/y 4) Sy 2 (6) where x is I(00), y is 1(450), and Sx and Sy are the standard deviations in x and y respectively. For 50% error in I(0') and 1(450), the propagated error is approximately 71% for 1(00)/1(450). The change in 1(00)/1(450) from a state of strong alignment to a state of no alignment is approximately three orders of magnitude, as 1(00)/I(450) changes from about 10-3 to 100. The error in a single data point is approximately 7% of that change. This clearly supports the conclusion that the transmission method is, in a real and meaningful way, able to distinguish aligned liquid crystals from unaligned liquid crystals. By taking many data points, the uncertainty in the alignment measurements was reduced. The alignment parameter A=I(00 )/I(450 ) was shown to be 1.04x10 -3 ± 2.8x10 -4 for brushed polyamide-imide aligned cells. The standard deviation in this measurement is about 27% of A, and only 3% of the change in A that occurs during the transition from the aligned to unaligned state. To determine whether the measured transition from the aligned to unaligned state is significant, a t-test is performed for the cells produced from softbaked polyamide-imide alignment films (Figure 5.15). For aligned cells with post-brush cure temperatures 2650 C and below, the average of the I(00)/1(450) values is 1.0x10- 3 ± 2.8x10 -4 . On the other 228 hand, for unaligned cells with post-brush cure temperatures 3000C and above, the average of the 1(00)/1(450) values is 0.98 ± 0.20. The quantity t can be calculated according to [37]: t = (x - y) / [(Sx 2/Nx) + (Sy2/Ny) 1/2 (7) where x and y are the two means compared, Sx and Sy are the standard deviations in x and y respectively, and Nx and Ny are number of measurements in x and y. For comparing the data sets of the aligned and unaligned specimens in Figure 5.13, the calculated value of t is about 8.6. According to tabulated critical values of t [37], t would have to be greater than 2.45 in order to conclude that there is a significant difference between these two data sets with 95% certainty, and above 5.96 in order to conclude that there is a significant difference between these two data sets with 99.9% certainty. Thus, clearly, with t=8.6, there is a significant difference between the aligned (post-brush cure <2650 C) and unaligned (post-brush cure 2300 0C) data sets with greater than 99.9% certainty. In conclusion, it is clear that the techniques described in this chapter provide a very real and meaningful measure of the parallel alignment of the liquid crystals. Error in the intensity measurements should not seriously affect the validity of the results. 5.5. Conclusions Brushed layers of Probimide 32 polyamide-imide parallel-aligned E7 nematic liquid crystals effectively. The main chains of the polyamide-imide are oriented during the brushing process. Also, 450A films of the preimidized Probimide 32 resin appear to be free of solvent after a 110 0 C cure. Polyimide, poly(amic acid), poly(vinyl alcohol), and poly(phenylene ether sulfone) also parallel-align E7 liquid crystals effectively. Empirically speaking, all of these polymers are yellow in color and have highly polar backbones. Polycarbonate and poly(methyl methacrylate) do not appear to produce parallel alignment of the liquid crystals. The "relaxation of alignment layers" has been introduced as a technique for evaluating thermal transitions in surface layers. When brushed polyamide-imide, PMDAAPB poly(amic acid), PMDA-APB polyimide, and poly(phenylene ether sulfone) layers are treated above their glass transition temperatures, their ability to align liquid crystals is irreversibly lost. This phenomenon has been taken advantage of in order to investigate processing / structure / property relationships in polymer surface layers. In addition, it was found that the ability of brushed layers of highly crystalline poly(vinyl alcohol) to align 229 liquid crystals was destroyed at its melting temperature, suggesting that polymer surface alignment can involve the orientation of either amorphous or crystalline chains, depending upon the polymer. 5.6. References 1. D. Berreman, Phys. Rev. Lett., 28(26), 1683 (1972). 2. D. Berreman, Mol. Cryst. Liq. Cryst., 23, 215 (1973). 3. U. Wolff, W. Greubel, and H. Kruger, Mol. Cryst. Liq. Cryst., 23, 187 (1973). 4. E. Lee, Y. Saito, and T. Uchida, Jpn. J. Appl. Phys., 32(12B), L1822 (1993). 5. Y. Zhu, L. Wang, Z. Lu, and Y. Wei, X. Chen, and J. Tang, Appl. Phys. Lett., 65(1), 49 (1994). 6. D. Seo, H. Matsuda, T. Oh-ide, and S. Kobayashi, Mol. Cryst. Liq. Cryst., 224, 13 (1993). 7. D. Seo, T. Oh-ide, H. Matsuda, T. Isogami, K. Muroi, Y. Yabe, and S. Kobayashi, Mol. Clyst. Liq. Clyst., 231, 95 (1993). 8. S. Ishihara, H. Wakemoto, K. Nakazima, Y. Matsuo, Liquid Crystals, 4(6), 669 (1989). 9. E. Lee, P. Vetter, T. Miyashita, T. Uchida, M. Kano, M. Abe, and K. Sugawara, Jpn. J. Appl. Phys., 32(10A), L1436 (1993). 10. Y. Kawata, K. Takatoh, M. Hasegawa, and M. Sakamoto, Liquid Crystals, 16(6), 1027 (1994). 11. H. Aoyama, Y. Yamazaki, N. Matsuura, H. Mada, and S. Kobayashi, Mol. Cryst. Liq. Cryst., 72, 127 (1981). 12. N. van Aerle and J. Tol, Macromol., 27, 6520 (1994). 230 13. J.M. Geary, J.W. Goodby, A.R. Kmetz, and J.S. Patel, J. Appl. Phys., 62(10), 4100 (1987). 14. E. Lee, P. Vetter, T. Miyashita, and T. Uchida, Jpn. J. Appl. Phys., 32, L1339 (1993). 15. S. Kuniyasu, H. Fukuro, S. Maeda, K. Nakara, M. Nitta, N. Ozaki, and S. Kobayashi, Jpn. J. Appl. Phys., 27(5), 827 (1988). 16. N. van Aerle, J. of the SID, 2(1), 41 (1994). 17. D. Seo, H. Matsuda, T. Oh-Ide, and S. Kobayashi, Mol. Cryst. Liq. Cryst., 224, 13 (1993). 18. N. van Arle, M. Barmentlo, and R. Hollering, J. Appl. Phys., 74(5), 3111 (1993). 19. T. Russell, MRS Bulletin, 21(1), 49 (1996). 20. M. Ree, T.L. Nunes, and D.P. Kirby, Polym. Prep., 33(1), 309 (1992). 21. B. Factor, T. Russell, and M.F. Toney, Phys. Rev. Lett., 66(9), 1181 (1991). 22. H. Fukuro and S. Kobayashi, Mol. Cryst. Liq. Cryst, 163 157 (1988).. 23. Y. Nakajima, K. Saito, M. Murata, and M. Uekita, Mol. Cryst. Liq. Cryst, 237 111 (1993). 24. D. Seo and S. Kobayashi, Appl. Phys. Lett., 66(10), 1202 (1995). 25. D. Seo, S. Kobayashi, and M. Nishikawa, Appl. Phys. Lett., 61(20), 1202 (1992). 26. M. Nishikawa, K. Sano, T. Miyamoto, Y. Yokoyama, M. Bessho, D. Seo, Y. Imura, and S. Kobayashi, Jpn. J. Appl. Phys., 33(7A), 4152 (1994). 231 27. S. Ishibashi, M. Hirayama, and T. Matsuura, Mol. Cryst. Liq. Cryst., 225, 99 (1993). 28. Amoco Ultradel 7100 series photosensitive polyimides, product information. 29. Personal communication, Dr. T. Gardner and Dr. R. Wenz, 3M Corporation. 30. J. Cognard, Mol. Cryst. Liq. Cryst., 78, Suppl. 1 (1982). 31. Merck E7 nematic liquid crystals, product information. 32. P.G. de Gennes, The Physics of Liquid Crystals, Ch. 3, Oxford Press, London, (1975). 33. C. Gooch and H. Tarry, J. Phys. D: Appl. Phys., 8, 1575 (1975). 34. Scientific Polymer Products, Inc., product information guide. 35. T.L. St. Clair, in Polyimides, p. 58, D. Wilson, H.D. Stenzenberger, and P.M. Hergenrother, Eds., Blackie, Glasgow (1990). 36. P. Cebe, Ph.D. Thesis, Cornell University (1984). 37. Laboratory Techniques Manual, Ch. 3, Department of Chemistry, Massachusetts Institute of Technology (1979). 232 Chapter 6. Alignment of Liquid Crystals on Photo-irradiated Polyamide-imide Layers. 6.1. Introduction In the previous chapter, the alignment of nematic liquid crystals on brushed polymer layers was discussed. As mentioned in Chapter 1, there is demand for techniques of alignment which do not require the surface to undergo a contact treatment. Techniques have been investigated to accomplish this, one of which is the exposure of a polymer layer to linearly polarized light. In this chapter, it is shown that the Probimide 32 polyamideimide can be processed as a photo-induced surface alignment material. Schadt et al. [1] previously showed that films of poly(vinyl methoxycinnamate) (PVMC) exposed to linearly polarized ultraviolet light were effective as surface alignment agents for nematic liquid crystals. The authors used a preferential crosslink argument to explain the phenomenon. One serious disadvantage of using a cinnamate polymer for alignment, however, is that it is a relatively obsolete photoresist material. It is not produced today in large, commercial quantities, nor does it sell at low cost. Hasegawa and Taira [2] recently showed that an aromatic/aliphatic polyimide produced liquid crystal surface alignment upon exposure of the films to linearly polarized short wave (X=257 nm) ultraviolet. This result was quite interesting because the polymer was not specifically designed for photosensitivity. The authors used a depolymerization argument, discussed in Chapter 1, to explain the alignment. One disadvantage of the technique was that the alignment was attained only in a very narrow range of exposure dosage, above and below which, no alignment was achieved. This could present difficulties, for example, in reproducing the alignment scheme should any processing conditions change. In this chapter, a technique of alignment which uses the photo-irradiation of polyamide-imide films is introduced. The photo-alignment process is compared to the rubbing process. Information is also sought about the mechanism of alignment. 6.2. Experimental Section 6.2.1. The photo-alignment technique The Probimide 32 polyamide-imide films and liquid crystal cells were produced as described in the experimental section of the previous chapter, section 5.2.1. The brushing 233 alignment was replaced here, however, with the linearly polarized ultraviolet (LP-UV) exposure process shown in Figure 6.1. The films were irradiated under a Spectroline Model R-51A short wavelength (X=254 nm) ultraviolet lamp. The UV was passed through an Ealing UV polarizer prior to incidence upon the bare alignment films. The films were irradiated at a constant two inch distance from the lamp. The area-normalized exposure rate was approximately lmW/cm 2. The alignment was evaluated by the qualitative approach and laser optical transmission technique discussed in section 5.2.2. UV Lamp X= 254nm I I I J i i i Linear to-Polarizer t I1 P C t Coatedl Slides l P r I i i I~ E , Figure 6.1. Exposure technique for irradiating alignment films with linearly polarized ultraviolet light (LP-UV). 234 6.2.2. Comments regarding the methodology by which PAI was selected It would be appropriate here to make some brief remarks regarding the way in which photo-alignment in the polyamide-imide was discovered. The Probimide 32 polyamide-imide was not selected as a candidate for photo-alignment studies with any foreknowledge of its ability to align liquid crystals. In fact, many non-contact alignment techniques were attempted before a successful technique was found. Producing liquid crystal alignment by exposure of short wavelength LP-UV to virtually every alignment material discussed in Chapter 5 was attempted. Exposure of inherently photosensitive polyimide materials, discussed in Chapters 1 and 3, to LP-UV of i-line wavelength, which crosslinks the material, was also tried. A non-contact, air jet brushing approach, and many variations of it, were also attempted. However, the only technique which produced the desired effect of inducing alignment was exposure of the polyamide-imide to 254 nm LPUV. In conclusion, the discovery of the photo-aligning properties in the Probimide 32 polyamide-imide was a culmination of an almost Edisonian style attack. Often, however, this is the only effective way of attaining a desired result. 6.3. Results and Discussion 6.3.1. Evidence of liquid crystal alignment upon LP-UV exposure Qualitatively, it was apparent that the LP-UV exposure to the polyamide-imide films induced parallel alignment of the E7 liquid crystals. Cells produced from LP-UV exposed polyamide-imide films exhibited a pattern of 900 extinctions consistent with an aligned material, while cells produced from unexposed and unbrushed films were bright, i.e., exhibited no extinctions, when placed between crossed polarizers. The alignment was therefore investigated by laser optical transmission. A liquid crystal cell with the alignment films irradiated with LP-UV for 25 minutes was rotated between the polarizer and analyzer with the alignment axis representing the 00 reference axis. For a given rotation angle 0, the optical transmission I(4) for a perfectly aligned system is expected to follow the equation: I(0)/I(450) = sin 2 (24) 235 which is plotted in Figure 1.13 in Chapter 1. In Figure 6.2, the normalized optical transmission I(4)/I(450) is plotted against 0 for the LP-UV aligned cell. The curve closely approximates the theoretical curve from Figure 1.13, thus indicating the the LP-UV irradiated polyamide-imide films effectively align the liquid crystals. The minimum I(0)/1(450 ) at the 00, 900, and 1800 extinction angles, though greater than the theoretical zero, is near 10- 3, which is comparable to brush aligned cells. The alignment parameter A=I(0 0 )/I(450 ) is plotted in Figure 6.3 as a function of LPUV exposure dosage to the polyamide-imide layers. For very small dosages A is near unity, indicating the absence of alignment. As exposure time increases, however, A declines, indicating that the LP-UV treatment to the polyamide-imide is inducing alignment in the liquid crystals. The alignment is very strong after about 25 minutes of exposure, where I(00)/I(450) is comparable to the 1(00)/1(450) = 10- 3 obtained in Chapter 5 for brush- aligned liquid crystal cells. The alignment remains strong for long exposure times. Hence, it is shown that the window of exposure time required to produce alignment is quite wide. .001 .0001 0 45 90 135 180 Angle (degrees) Figure 6.2. I(0)/1(450 ) vs. rotation angle for liquid crystal alignment cell produced with polyamide-imide alignment layers that have been irradiated with LP-UV for 25 minutes. 236 10" .01 I .001 .001 () .0001 44 - 60 120 180 240 300 360 420 480 LP-UV Exposure Time (Minutes) Figure 6.3. 1(00)/1(450) vs. LP-UV exposure time for liquid crystal cells produced with LP-UV irradiated polyamide-imide alignment films. 237 6.3.2. Effects of unpolarized UV exposure Fourier transform infrared spectroscopy It was desired to understand how the ultraviolet irradiation affects the polyamideimide alignment film. FTIR spectra of unexposed and UV exposed polyamide-imide films are shown in Figure 6.4. The region in which changes can clearly be seen is in the 12001900 cm- 1 range. The unexposed polyamide-imide film (Figure 6.4(a)) was irradiated with unpolarized UV for one hour with an area-normalized exposure rate of approximate 7mW/cm 2 (Figure 6.4(b)). The most apparent change is the reduction in C=C phenyl stretching near 1500 cm- 1 which declines about 50% in peak height relative to C=O stretching at 1720 cm- 1 from the unexposed to exposed samples. Also, the N-C imide vibration at 1370 cm- 1 declines in height about 25% relative to the 1720 cm- 1 peak. In addition, the amide carbonyl band at 1680 cm- 1 recedes into a shoulder upon UV exposure. These changes suggest that there are bond degradation reactions occurring in the polymer upon exposure to UV with the phenyl rings the primary site. Alignment layer relaxation Upon exposing a polymeric material to extreme conditions, such as to high temperatures or to irradiation, bond breaking, and possibly bond reformations, often occur. The net effect on structure and properties will often depend upon whether crosslinking or chain scission dominate the process. In Chapter 4, for example, it was found that upon heating the bulk films to high temperatures in air, chemical changes lead to a net increase in molecular weight. This was indicated by an increase in Tg and reinforcement of the rubbery modulus that was apparent in the bulk dynamic mechanical spectra. The effect of UV on the alignment films is difficult to measure, however, via conventional bulk film measurements . For spin coated films 41tm thick, several hours of UV lamp exposure has no apparent effect on the FTIR spectra of the material. Kambe [3] reports that very long UV lamp exposure times, on the order of hundreds of hours, are required to degrade the bulk mechanical properties of aromatic polyimides, polyamides, and polyamide-imides. In any case, it may not be reasonable to assume that the effects of long term exposure on bulk films are similar to the changes that occur in the surface layers upon shorter exposures. 238 0 0 0 rM E .0 QD T- 0 D C14 (sl!un Aej~aqabe) eoueqjosqv a Iojela2p Figure 6.4. Relative absorbance vs. wavenumber for a Probimide 32 polyamide-imide film. a.) unexposed to UV, and 239 O -- -- ---------------------~~ C C -o o a E = CP o C C) CI C·1 - ~~~ -~---~I-~ (si!un AJe4l!qJe) eoueqJosqV eA!IleleL Figure 6.4. continued. b.) exposed to UV. 0 240 Fortunately, a technique called alignment layer relaxation, through which the thermal stability of surface layers can be investigated, was introduced in Chapter 5. This method was used here to evaluate the changes in thermal stability of UV irradiated polyamide-imide surfaces. Polyamide-imide alignment films on glass were exposed to unpolarized UV for 2 hours. The films were brushed with rayon, and consequently cured at an array of temperatures. Liquid crystal cells were built, and 1(00)/1(450) for the cells was plotted against post-brush cure temperature. The results, shown in Figure 6.5, indicate that the UV exposure reduces the Tg of the surface layers. This suggests that chain scission reactions dominate upon exposure to short wavelength UV. This furthermore suggests that the photo-aligning process in the polyamide-imide is associated with depolymerization rather than with crosslinking. a .1 .001 .0001 - - I 100 I 150 S 200 I 250 I 300 II 350 400 Temperature (oC) Figure 6.5 1(00)/1(450) vs. post-brush curing temperature for liquid crystal cells produced with UV exposed (squares) and unexposed (circles) polyamide-imide alignment layers. 241 6.3.3. Effects of linearly polarized UV exposure Fourier transform infrared spectroscopy Linearly polarized UV and unpolarized UV appear to affect the polyamide-imide in a similar way. Furthermore, dichroic FTIR spectra of LP-UV exposed material in Figure 6.5, in contrast with brushed layers, do not indicate the development of anisotropy. The linearly polarized UV appears to reduce the height of the absorption at 1500 cm- 1 after 2 hours of LP-UV exposure (Figure 6.6(a,b)), and even more after 4 hours of LP-UV exposure (Figure 6.6(c,d)) This change was also observed previously as a result of the unpolarized UV treatment. FTIR spectra taken with the incident IR axis polarized parallel to the LP-UV axis (Figure 6.6(a,c)) appear to be identical to those taken with the incident IR perpendicular to the LP-UV axis (Figure 6.6(b,d)). This is dissimilar to dichroic FTIR data for the brushed polyamide-imide which indicated that the main chains were oriented in the direction of the brushing. These data of the LP-UV exposed sample suggest that the linearly polarized UV, unlike the brushing, does not physically move the polymer chains with respect to the LPUV axis. It is also possible that the dichroism produced is too small to detect by conventional FTIR. This could be the case, for example, if orientation exists in only the first monolayer. Because the mechanism of liquid crystal alignment on polymer surfaces is not entirely understood, it is difficult to speculate as to what exactly is the mechanism of the UV-induced surface alignment for the polyamide-imide. Alignment layer relaxation The alignment layer relaxation method was applied for liquid crystal cells aligned with unbrushed, but LP-UV exposed, polyamide-imide layers. Many films were exposed to 2 hours of LP-UV, a treatment which was shown previously (see Figure 6.3) to produce strong liquid crystal alignment. The films were cured at an array of temperatures, and liquid crystal cells were built. The value of I(00)/I(450) is plotted against post-UValignment curing temperature in Figure 6.7. It is apparent that the LP-UV exposed layers do in fact show a thermal transition. This result indicates that, though ordering in the LPUV exposed alignment layers was not detected by FTIR, the relaxation of the film surface reverses the liquid crystal aligning properties introduced by the LP-UV. It is also apparent that chain scission caused by exposure to the LP-UV reduces the glass transition temperature of the polyamide-imide surface layer. 242 o o E o ° O o E >) -o C 1r 0r (sl!un Ajeal!qje) eoueqJosqV eA!•eleH Figure 6.6. Relative absorbance vs. wavenumber for polyamide-imide layers exposed to linearly polarized ultraviolet (X=245 nm). a.) 2 hour exposure, UV axis parallel to incident IR axis, 243 0 C) 0 0 o C) CN 0 C) o 2 0 O C) 2 = E C) = > (sl!un •iedj!qje) eoueqjosqV a!jeledl Figure 6.6. continued. b.) 2 hour exposure, UV axis perpendicular to incident IR axis, 244 0 0 0 0 SE0 a r o0 o E C) 0) 0 0 0 0o (sl!un AjeJl!qje) aoueqJosqv a^!elae= Figure 6.6. continued. c.) 4 hour exposure, UV axis parallel to incident IR axis, 245 0 0 0 0 0 CD~ CD 0 o O E C.D Lm E0 E 6 (silun Aaedqaed ) aoueqjosqV aiA!elae Figure 6.6. continued. d.) 4 hour exposure, UV axis perpendicular to incident IR axis. 246 ,, A A 1I o001.0 .001 1 .0001 - 50 I 100 150 I 200 I 250 I 300 I 350 400 Temperature (oC) Figure 6.7 I(0O)/I(450) vs. post-aligning cure temperature for liquid crystals cells produced with polyamide-imide alignment films that have been exposed to LP-UV (triangles) or brushed (circles). Atomic Force Microscopy An atomic force microscopy (AFM) image of a Probimide 32 polyamide-imide film irradiated 2 hours with LP-UV is shown in Figure 6.8. There appear to be no microscopic effects of the UV when compared to the AFM of pristine material shown in Chapter 5, Figure 5.13(b). This is in contrast to the brushed films, shown in Figure 5.13(a), which had a rough surface and grooves running in the direction of the brushing. 247 N N 0, 0 0 U, zC 0 N Cq w Figure 6.8. Atomic force microscopy (AFM) image of LP-UV irradiated polyamide-imide films. 248 UV stability of the alignment The UV stability of the liquid crystal alignment was tested by the following procedure: 1. Many polyamide-imide films were treated with an alignment inducing process, i.e., brushing or LP-UV exposure. 2. The films were then exposed to unpolarized UV, approximately 7mW/cm 2, for an array of exposure times, at least two films per exposure time. 3. Liquid crystal cells were constructed. 4. 1(00)/1(450) was measured and plotted vs. post-aligning UV dosage. Measured values of 1(00)/1(450) are plotted against post-aligning UV dosage in Figure 6.9. The brushed films did not lose much of their aligning properties for any postbrush UV exposure. This is most likely because the alignment in the brushing alignment process is associated with the physical movement of main chains in the alignment film. Exposure to UV was apparently not sufficient to destroy the aligning characteristics of the film. For the LP-UV aligned layers, however, subsequent unpolarized UV exposure appears to damage the ability of the polyamide-imide films to align the liquid crystals. This experiment indicates that the LP-UV does not align the liquid crystals via the physical movement of main chains. If that were the case, the unpolarized UV should not be able to reverse the alignment, as it did not in the case of brushed films. Alternatively, if the LPUV exposure produces alignment by scissioning the polymer preferentially along one axis, it is possible that further exposure of the material to unpolarized UV might act to deplete the groups that are otherwise unaffected by the LP-UV, thus mitigating the aligning impact of the LP-UV. 249 II O 0 0O .001 i .0001 - o 0e 0 I i1 120 180 UV Exposure Time (Minutes) Figure 6.9. 1(00)/1(450) vs. post-aligning unpolarized UV exposure time for liquid crystal cells produced from polyamide-imide alignment films that were exposed to LP-UV (open circles) or brushed (filled circles). 250 6.4. Conclusions The parallel alignment of nematic liquid crystals has been accomplished by use of a polyamide-imide alignment film that has been exposed to linearly polarized short wavelength ultraviolet (X= 254 nm). Alignment was produced within minutes of exposure and was maintained with very long exposures. Alignment layer relaxation and Fourier transform infrared spectroscopy studies suggest that the short wavelength UV causes chain scission in the polyamide-imide, though it is difficult to speculate at this time as to the mechanism of alignment. Unlike the brushing technique, neither macroscopic grooves nor bulk orientation of the main chains are produced in the polymer films by the LP-UV. While brushing-induced alignment appears to be stable upon exposure to unpolarized short wavelength UV, the UV-induced alignment appears to be damaged by subsequent exposure of the bare alignment films to unpolarized short wavelength UV. 6.5. Subjects for further work It was the goal of this research to demonstrate a new way of aligning liquid crystals and to gain some better understanding into the science of liquid crystal alignment. Though these goals were accomplished, this research has opened doors to topics for further investigation. They are listed below, and are divided between the practical and scientific questions. It will be quite interesting to see whether or not, over the next 5-10 years, these questions are answered. Can effective commercial devices be produced in an economical way via a photo-alignment process? This raises many questions. 1. Can adequate surface pretilt be produced by a photo-aligning technique? And how might one go about producing surface pretilt by a photo-aligning process? Might alkyl side groups added to the repeat unit of the polymer or oblique irradiation of the alignment film produce pretilt? 2. Can the speed of the photo-alignment process be improved such that it is comparable to brushing? And how might one go about speeding it up? Will increasing the intensity or adding a radical initiator work? 251 3. Can a commercial display with adequate switching speed and device lifetime be manufactured via a photo-alignment technique? 4. If other polymers can be used in photo-alignment processes, how do they compare to the polyamide-imide in terms of the alignment strength produced and the window of exposure dosage required? 5. Will the photo-alignment work with other types of nematic liquid crystals or in other liquid crystal technologies? Can it be used to align smectic or ferroelectric liquid crystals? Can the photo-alignment approach be used in polymer dispersed liquid crystal displays? What is the mechanism of photo-alignment? This is probably a more difficult question which probably also requires more insight into how the brushing-induced alignment works. This raises the following questions: 1. What types of polymers produce liquid crystal alignment by the brushing process, and which ones do not? Why? 2. What polymers have liquid crystal photo-aligning properties? What, then, are the characteristics of these polymers? 3. How do liquid crystals interact with brushed polymer surfaces, and how does this compare with the way they interact with LP-UV exposed polymer surfaces? Can this be investigated by molecular modeling, second harmonic generation experiments, or other surface sensitive techniques? 4. If brush-induced alignment has been suggested to be associated with crystallization, is it possible that the UV produces surface crystallization? 5. Though the dichroic FTIR experiments could not be detect ansiotropy in LP-UV exposed films, might there be other techniques that can? Is the orientation too small to detect? Does it occur in only the first monolayer? Or might there be no dipolarity at all? 252 6.6. References 1. M. Schadt, K. Schmitt, V. Kozinkov, and V. Chigrinov, Jpn. J. Appl. Phys., 31(7), 2155 (1992). 2. M. Hasegawa and Y. Taira, Rec. SID Int. Disp. Res. Conf., 213 (1994). 3. H. Kambe, in Aspects of Degradationand Stabilization of Polymers, Ch. 8, H. Jellinek, Ed., Elsevier, Amsterdam (1978). 253 Appendix A. Description of Nematic Alignment in Terms of the Poincare Sphere A.1. The Poincare sphere The Poincare sphere [1,2] is a geometrical construction upon which any polarization state can be described. Every latitudinal / longitudinal coordinate represents a different polarization state. As shown in Figure A.1, the lines of latitude describe the handedness and ellipticity (1) of the electric field vector. Light in the northern hemisphere is left handed, while light in the southern hemisphere is right handed. The closer to the equator, the lower is p. At the equator, 1 = 0, and the light is linearly polarized. At the poles, 11 = 1, the north pole representing left circularly polarized light, and the south pole representing right circularly polarized light. The equator is defined as 00 latitude, the south pole +900 latitude, the north pole -900 latitude. The ellipticity P is related to latitude (I) by equation (1). I = 2 arctan p (1) -I 0=o +1 Figure A. 1. Lines of latitude (I) on the Poincare sphere representing ellipticity ([3). 254 As shown in Figure A.2, the lines of longitude represent different azimuthal angles. The prime meridian, 00 longitude, is the point where the azimuthal angle is 00, i.e., parallel to the axis of reference. The azimuth (a) increases as we travel clockwise from 00 longitude, as described in equation (2), where L represents degrees longitude. L=2 a (2) Thus, two points along the equator separated by 1800 longitude represent two linearly crossed polarized states. 1800 Figure A.2. Lines of longitude (L) on the Poincare sphere representing azimuth (a). A.2. Changes in polarization state on the Poincare sphere As light passes through a substance, the state of polarization of the light can change. Such a change can be represented by a relative positional change along the surface of the Poincare sphere. If the light transverses multiple layers, the construction can be repeated as many times as needed to determine the final polarization state. We begin at a given point on the sphere. Movement on the sphere proceeds, in general, along an appropriate circle on the surface of the sphere. (Equivalently, we can fix the initial point and rotate the sphere along an appropriate axis.) 255 The polarization state of the light passing through the polarizer is quite easy to find on the Poincare sphere. The emerging light is linearly polarized, and thus, rests on the equator. For convenience, we can define this as zero azimuth, and place the initial state at 00 latitude and 00 longitude. The change in polarization state as the light passes through a birefringent sample, i.e., an aligned LC film, is found as shown in Figure A.3. First, we locate the point P which represents the polarization state of the light before it enters the sample, that is, linearly polarized light. Having chosen point P as our point of reference, movement along the sphere proceeds as follows. We locate the point S which represents the fast vibration axis of the retarding material. In the parallel aligned liquid crystal cell, this would be the location on the equator which is perpendicular to the nematic director. Then, we draw the small circle on the surface of the Poincare sphere which intersects P and is normal to OS, the radial vector through S. The polarization state of the light after it has passed through the sample is found by traveling from P clockwise (looking from S) along the circle an arc length equal to 8 given in equation (24) of Chapter 1. We can, of course, produce the same result by fixing the initial point in space and rotating the sphere about OS. Figure A.3. Movement along the Poincare sphere from reference point P along the small circle of the sphere centered on OS, where O is the center of the sphere and S represents the azimuthal angle parallel to the fast vibration axis of the crystal. 256 As shown in Figure A.4, the change in polarization state as the light passes through the analyzer is found by moving along the small circle on the surface of the sphere which is in the plane normal to the pole-to-pole axis. In other words, we make purely a longitudinal shift along a line of latitude. The arc length moved is twice the azimuthal angle of the analyzer. We can, equivalently, fix the initial point and rotate the sphere about the polar axis. Figure A.4. The relative change in polarization state induced by the analyzer is a longitudinal shift through the arc length between the analyzer (A) and the point of reference (P). X1 moves to X2. In Figure A.5, we suppose the position on the Poincare sphere to have moved from P1l, position of the polarizer, to P2, after having passed through the sample, and then to P3, after passing through the analyzer. The intensity of the light leaving the analyzer I/E 2 is shown by Jerrard [2] to be given by the simple expression in equation (2). I/E 2 = cos 2 p (2) In equation (2), 2p is the arc length between the reference point (P1) and final point (P3). 257 Figure A.5. The arc length from reference state P1 to final state P3 is given by 2p. The intensity transmitted by the system is cos 2p where full intensity is unity. A.3. Examples of LC nematic orientation Figure A.6 depicts the case in which the polarizer and analyzer are crossed, and the liquid crystal plate is oriented with the nematic vector parallel to the polarizer. Points P, A, and S represent the axes of the polarizer, analyzer, and fast vibration axis respectively. There is no change in polarization state as the light passes through the sample because S is diametrically opposed to P. Because the polarization state remains diametrically opposed to A, position of the analyzer, no light is transmitted. In other words, the analyzer in azimuth 900 imposes a 1800 longitudinal shift. The arc length rtrepresents 2p, and by equation (2), the intensity transmitted by the analyzer is zero, i.e. cos 2(n/2). Figure A.7 depicts the case in which the nematic vector is 450 with respect to the polarizer. While the circle in Figure A.7 represents the equator, the line though the middle represents the prime meridian. As light from the polarizer is transmitted through the sample, the polarization state of the light moves from the linearly polarized state at point P to the elliptically polarized state at point E. In this geometry, light is permitted to pass with the intensity given by equation (2). Note that a 450 sample orientation has the unique property of imposing no change in the azimuth of the light ray. 258 S, A Figure A.6. Parallel aligned LC plate with nematic vector parallel to P. Figure A.7. LC cell with nematic vector 45' with respect to P and A. The light is elliptically polarized by the sample as it moves along the prime meridian to point E. 259 A.4. Effect of misalignment of the liquid crystals Figure A.8 demonstrates the effect of inhomogenous phases on the polarization state of light as represented on the Poincare sphere. In Figure A.8, the alignment director is parallel to the polarizer. In an ideally aligned system, the final polarization state will be a single point A with reference to end point arc length 2p = 1800 (and cos2 p = 0). Unique inhomogenous phases, on the other hand, will create a distribution of polarization states on the surface of the sphere deviating from the single point A. Hence, the observed intensity will be greater than cos 2 (900). Figure A.8. Effect of random inhomogenous region on the Poincare sphere analysis. The shaded region represents a distribution of polarization states deviating from A. A.5. References 1. C. Viney, TransmittedPolarisedLight Microscopy, McRone Research Institute, Chicago (1990). 2. H. Jerrard, J. Opt. Soc. Am., 44(8), 634 (1954).